Mellanox Technologies MC2210130-003 Datasheet

Brand: Mellanox Technologies

Model: MC2210130-003

Type: Datasheet

This manual is also suitable for

July 2011

Corporate Style and Branding Guide

CORPORATE STYLE GUIDE

2 | Mellanox Technologies | July 2011

PAGE 1

LOGOTYPE

SIGNATURE

GRAYSCALE LOGO

Corporate Logo Usage

CORPORATE LOGO/SIGNATURE

The Mellanox Technologies signature is used to

represent the organization in all its activities. This

signature is the foundation to our identity and exists on

a wide variety of media and marketing materials.

The Signature consists of two elements:

1. the symbol

2. the logotype

Because the signature is a registered trademark, the

relationship between these elements should never be

altered. This ensure legal protection, builds recognition

and reinforces our positioning.

CORPORATE 2-COLOR LOGO/SIGNATURE

PMS 274 Blue

60% Black

100% Black

60% Black

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 3

CMYK CONVERSION

COLOR CYAN MAGENTA YELLOW BLACK

Blue 100 100 0 28

Gray 0 0 0 60

RGB CONVERSION

COLOR RED GREEN BLUE

Blue 34 31 114

Gray 130 130 130

WEB CONVERSION

COLOR RED GREEN BLUE

Blue 22 1F 72

Gray 82 82 82

LOGO VARIATIONS

Corporate Logo Usage

These acceptable variations of the corporate logo

should only be used when space or application dictate

its use.

The horizontal version is used as an EXCEPTION.

CORPORATE STYLE GUIDE

4 | Mellanox Technologies | July 2011

LOGO CLEAR SPACE

LOGO MINIMUM SIZE

No smaller than 1”

No smaller than 1.5”

STACKED LOGO

HORIZONTAL LOGO

Logo Clear Space and Size

The signature should always be surrounded by an adequate amount of

clear space in order to set it off from other elements.

The gray area (see illustration at left) indicates the minimum amount of

clear space that must surround the signature in all applications. No other

elements should infringe in the clear space.

Exceptions require approval prior to use.

Minimum clear space is specied in units of “X.” X equals the height of

the “x” in Mellanox.

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 5

LOGO USAGE GUIDELINES

This is the ofcial corporate logo.

The Logo should be used as a single unit. The logotype should

never be used alone, but should always have the logo mark

(bridge).

The signature or logotype should NEVER by used separately.

Do not rearrange or stack the logo mark and

logotype.

Do not change the colors of the logo or logotype.

Logo Usage

CORPORATE STYLE GUIDE

6 | Mellanox Technologies | July 2011

Do not tilt or skew logo.

Do not enlarge or shrink the logo or the logotype

separately.

Do not place logo on a background or backgrounds

with conicting colors.

White is the recommended background.

LOGO USAGE GUIDELINES

Logo Usage

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 7

PRODUCT/TECHNOLOGY LOGOS

OTHER MELLANOX LOGOS

Secondary and Product Logos

Mellanox product or secondary logos should also be used with the

same requirements as the corporate logo.

• They should always be surrounded by an adequate amount of

clear space in order to set it off from other elements.

• The Logo should be used as a single unit.

• Do not rearrange or stack the logo mark and logotype.

• Do not change the colors of the logo or logotype.

• Do not tilt or skew logo.

• Do not enlarge or shrink the logo or the logotype separately.

• Do not place logo on a background or backgrounds with

conicting colors.

• White is the recommended background.

CORPORATE STYLE GUIDE

8 | Mellanox Technologies | July 2011

FONTS

LEGACY FONTS

Bell Gothic Std LIGHT

Bell Gothic Std Medium

Bank Gothic BT LIGHT

Bank Gothic BT Medium

Bell Gothic Std typeface is primarily used for Speciality

Headlines and Subheads larger than 11 pt.

Bell Gothic should not be used for body copy or type

smaller than 12 pt unless used in ALL CAPS

Use -30 Horizontal tracking/kerning

Bank Gothic BT typeface can be used for Speciality

Headlines and Subheads larger than 14 pt. This is a legacy

fonts that has been used prior to 2010, and has been

replaced by Bell Gothic Std. but it is still acceptable to

use.

Bank Gothic should never be used for body copy because

of it’s ALL CAPS nature.

BELL GOTHIC STD LIGHT

BELL GOTHIC STD MEDIUM

Bank Gothic BT LIGHT

Bank Gothic BT Medium

Univers LT 39 Thin Ultra Condensed

Univers LT 45 Light

Univers LT 45 Light Oblique

Univers LT 47 Condensed Light

Univers LT 47 Condensed Light Oblique

Univers LT 53 Extended

Univers LT 53 Extended Oblique

Univers LT 55 Roman

Univers LT 55 Oblique

Univers LT 57 Condensed

Univers LT 57 Condensed Oblique

Univers LT 59 Ultra Condensed

Typography

Univers LT (True Type Version) is the primary fonts used on all corporate materials.

Arial or Helvetica may be substituted for Univers for online applications and PowerPoint

Univers LT 63 Bold Extended

Univers LT 63 Bold Extended Oblique

Univers LT 65 Bold

Univers LT 65 Bold Oblique

Univers LT 67 Bold Condensed

Univers LT 67 Bold Condensed Oblique

Univers LT 73 Black Extended

Univers LT 73 Black Extended Oblique

Univers LT 75 Black

Univers LT 75 Black Oblique

Univers LT 85 Extra Black

Univers LT 85 Extra Black Oblique

Univers LT 93 Extra Black Extended

Univers LT 95 Extra Black Extended

Oblique

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 9

SECONDARY CORPORATE COLORS

COLOR CYAN MAGENTA YELLOW BLACK

PMS 138 GOLD 0 30 100 0

30% BLACK 0 0 0 30

COLOR RED GREEN BLUE

PMS 138 GOLD 253 185 19

30% BLACK 190 190 190

Color Palette

SUGGESTED COLOR COMBINATIONS - 3 COLOR

SUGGESTED COLOR COMBINATIONS - 4 COLOR

OTHER AVAILABLE COLORS

COLOR CYAN MAGENTA YELLOW BLACK

PMS 282 BLUE 100 68 0 55

PMS 282 BLUE - 60% 60 40 0 32

PMS 371 GREEN 43 0 100 56

PMS 159 ORANGE 0 66 100 7

PMS 284 LIGHT BLUE 55 19 0 0

PMS 7545 BLUE GRAY 23 2 0 63

COLOR RED GREEN BLUE

PMS 282 BLUE 0 45 98

PMS 282 BLUE - 60% 80 102 147

PMS 371 GREEN 79 111 25

PMS 159 ORANGE 227 111 30

PMS 284 LIGHT BLUE 108 173 223

PMS 7545 BLUE GRAY 92 111 124

PMS

274

PMS

274

60%

BLACK

60%

BLACK

PMS

282

PMS

282

PMS

138

PMS

138

60%

PMS

282

PMS

282

60%

PMS

282

PMS

274

PMS

274

60%

BLACK

60%

BLACK

PMS

371

PMS

284

PMS

284

60%

PMS

282

PMS

274

PMS

274

PMS

282

PMS

284

PMS

159

PMS

159

PMS

371

PMS

371

PMS

7545

CORPORATE STYLE GUIDE

10 | Mellanox Technologies | July 2011

A trademark is a word, phrase, logo, symbol

or design, or a combination of these elements,

used to identify or distinguish the goods and

services of one company or individual from

others.

Using the trademark properly is necessary

in order to demonstrate that a mark is

used in commerce, which is a fundamental

requirement for trademark ownership in the

United States.

TRADEMARKS

FabricIT

MLNX-OS

SwitchX

REGISTERED TRADEMARKS

Mellanox

BridgeX

ConnectX

CORE-Direct

InniBridge

InniHost

InniScale

PhyX

Virtual Protocol Interconnect

Voltaire

Trademarks, Registered Trademarks and Copyrights

Always use the mark as an adjective

It is very important to always use the marks as adjectives and

always with the generic terms which they modify. A trademark

should never be use it as a noun or a verb.

INCORRECT:

ConnectX

is the best adapter on the market..

CORRECT USE:

ConnectX

adapters are the best on the market.

Avoid plural or possessive forms of the mark

Never use a mark in the plural form or the possessive form.

Never hyphenate trademarks

INCORRECT:

ConnectX

-compatible adapters are the best on the market.

CORRECT USE:

“ConnectX

compatible adapters are the best on the market.

Appropriate placement of trademarks within text

Generally, the mark must be used with the rst or most prominent

appearance of a trademark in a publication or document, but need

not be used with each subsequent appearance. As a safeguard,

use additional markings rather than fewer within a document.

Trademark Attribution Statement

For publications containing third party trademarks, it is typical

practice to provide a trademark attribution statement in small print

at the end of the specic article. Below is the trademark statement

that should be on all Mellanox marketing collateral.

Mellanox, BridgeX, ConnectX, CORE-Direct, InniBridge, InniHost,

InniScale, PhyX, Virtual Protocol Interconnect and Voltaire are

registered trademarks of Mellanox Technologies, Ltd. FabricIT,

MLNX-OS and SwitchX are trademarks of Mellanox Technologies,

Ltd. All other trademarks are property of their respective owners.

The notice should always contain:

1. The symbol © (the letter C in a circle), or the word “Copyright”;

2. The year of rst publication of the work; and

3. The name of the owner of copyright in the work.

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 11

PRINT COLLATERAL

Mellanox has many types of marketing collateral, signage, multimedia

and technical documents. All collateral should follow corporate branding

guidelines, colors, fonts and established templates when available.

Detailed specications for the Product Briefs and Case Studies and can be

found on starting page 13. All documents follow similar layouts, fonts

and specications. Samples without detailed specications only indicated

attributes that are different or unique for that document.

Some of the standard document element are in an InDesign library.

All standard documents should be created and maintained in Adobe InDesign

unless otherwise indicated.

DOCUMENT NUMBERS

The following documents should always contain a document number on the

last page in the lower right corner. They should follow this format:

Product Briefs = XXXXPB Rev 1.0

Solution Brief’s = XXXXSB Rev 1.0

White Paper = XXXXWP Rev 1.0

Case Studies = XXXXCS Rev 1.0

Data Sheets= XXXXDS Rev 1.0

ILLUSTRATIONS AND MULTIMEDIA

Mellanox has developed illustrations and icons to be used in PPT and other

collateral. In order to maintain consistency in imagery it is important to use the

established illustrations and icons whenever possible.

Illustration and icon examples can be found on starting on page 22.

Mellanox also utilizes Flash, animation and video for marketing purposes. It is

also recommended that the corporate branding, colors and fonts be used as

much as possible.

Documents and Marketing Collateral

The following documents have

pre-established templates:

> Product Briefs

> Solution Briefs

> Case Studies

> White Papers

> Datasheets

> Technology Sheets

> Reference Guides

> Product Brochures

!"#"$%&&#



>)0'%"

!"#$%&'()&*$+&,'-&./$0123)$4##/$0155.6&7)/$89$:;#<"





!%G+>S510/6%"#$$3%&'(()*+,%-'./*+(+01'23%4((%510/62%5'2'57'83

!'')*25,% #!5*))!.5%,!.5)4)%'-.5)4)%,% #!5)4)%*-.5)4)%5)4)%'!5)4)%5$35%,./'5,*.**'5).!,*))!.5) 5*'.%,!5,!5,!#%-.!,! 5

., !(,&-5*"5!'')*25!$)*'*#%!-5. 5,%5%-55., !(,&5*"5!'')*25!$)*'*#%!-5. 5''5*.$!,5., !(,&-5,!5+,*+!,.35*"5.$!%,5,!-+!.%0!5*1)!,-

="55"64;'E6FG5%"4$G6

$!5!'')*25--%0!5*++!,5%#55'!-5%)'/ !553!,5

(1L16'8%/)58A)5'%A)55)*6S\%A/1./%.+7'52%5'>)15%+5%5'>().'L'*63

:88$4$G6"2'E6FG5%"4$G6

*,5(*,!5%)"*,(.%*)5*/.5!'')*255'!5-*'/.%*)-5+'!-!5

.+*6).6%S+@5%&'(()*+,%-'./*+(+01'2%2)('2%5'>5'2'*6)617'%+5%71216]%

/66>]OOAAA3L'(()*+,3.+L^[email protected]^_G)?('2^_G)?('2%K7'571'A

5!05

!"#$%&'()'

!"#$%&'*)'

CORPORATE STYLE GUIDE

12 | Mellanox Technologies | July 2011

Documents and Marketing Collateral

COLOR DESIGNATIONS

Some of the Mellanox documentation and marketing collateral follow a color convention for specic product families or

technologies. Below is the color designation and the product area associated with the color.

Mellanox Product Briefs and Product Brochures follow the color convention.

PMS 282

Switches, Gateways,

Switch Silicon, Gateway Silicon, Phy Silicon

PMS 371

Ethernet and InniBand Adapter Cards,

Ethernet Adapter Silicon

PMS 159

Software

PMS 7545

Cables, Modules

PRODUCT BRIEF

350 Oakmead Parkway, Suite 100, Sunnyvale, CA 94085

Tel: 408-970-3400 • Fax: 408-970-3403

www.mellanox.com

Mellanox, BridgeX, ConnectX, CORE-Direct, InniBlast, InniBridge, InniHost, InniRISC, InniScale, InniPCI, PhyX, Virtual Protocol Interconnect and Voltaire are registered trademarks of Mellanox Technologies, Ltd.

FabricIT is a trademark of Mellanox Technologies, Ltd. All other trademarks are property of their respective owners.

CABLES

HIGHLIGHTS

Overview

Mellanox active optical modules provide outstanding performance for high-bandwidth fabrics, extending

the benets of Mellanox’s high-performance InniBand and 10 Gigabit Ethernet adapters throughout the

network. In addition to meeting or exceeding IBTA and IEEE standards, Mellanox certied modules are

100% tested on Mellanox equipment to ensure optimal signal integrity and the best end-to-end perfor-

mance. Mellanox QSFP module uses MTP/MPO connector type to connect an OM3/OM4 ber cable.

Mellanox SFP+ modules use LC connector type to connect to single mode or multi mode ber cable.

Mellanox optical solutions provide short, medium and long range scalability for all topologies, utilizing

innovative optical technologies to enable extremely high signal integrity and reliability. Rigorous produc-

tion testing ensures best out-of-the-box installation experience, performance and durability.

Table 1 - Module Specications and Ordering Information

Form Factor Targeted Application Max Power Consumption [W] Data Rate [Gb/s]

Min | Max

Part Number

QSFP 40Gb/s, QDR, InﬁniBand 1.7 2.5 10.0 MFM4R12C-MFM

SFP/SFP+ 10GigE, 10GBase-SR 1 9.95 10.5 MFM1T02A-SR

SFP/SFP+ 10GigE, 10GBase-LR 1 9.95 10.5 MFM1T02A-LR

Table 2 - Absolute Maximum Ratings

Parameter Range

Supply Voltage 3.45V

Operating Case Temperature 0°C to 70°C

Storage Ambient Temperature -10°C to 75°C

Table 3 - Operating Conditions

Parameter Minimum Typical Maximum Units

Supply Voltage 3.15 3.3 3.45 V

Input Voltage -0.3 3.6 V

Humidity (RH) 5 85 %

Active Optical Modules

Warranty Information

The Mellanox optical modules include a 1-year limited hardware warranty, which covers parts repair or

replacement.

Additional Information

For more information about Mellanox cable solutions, please contact your Mellanox Technologies sales

representative or visit: http://www.mellanox.com=>Products=>Cables=>Cables Overview

– Low power consumption

– Compliant with SFF-8436

– Compliant with SFF-8431

– Compliant with SFF-8432

– Low insertion loss

– BER Better than 10

-15

– Tested in Mellanox’s end-to-end

environment

– RoHS 6 compliant

PRODUCT BRIEF

ETHERNET

ADAPTER CARDS

ConnectX-2 EN 40 Gigabit Ethernet Network Interface Card (NIC) delivers

breakthrough connectivity for servers and storage. IT managers will be

able to build out high-bandwidth fabrics for next generation data centers

enabling the highest platform efciency and CPU utilization. Virtualized

environments will benet as well from the hardware-based accelerations

enabling more virtual machines per server and superior application

performance.

The single port ConnectX-2 EN 40G adapter

comes with a QSFP connector suitable for use

with copper or ber optic cables.

World-Class Ethernet Performance

ConnectX-2 EN 40G with PCI Express 2.0 delivers

industry-leading Ethernet bandwidth required by

multi-core, multi-socket platforms. Hardware-

based stateless ofoad engines handle the TCP/

UDP/IP segmentation, reassembly, and checksum

calculations that would otherwise burden the

host. Total cost of ownership is optimized by

maintaining an end-to-end Ethernet network on

existing operating systems and applications.

Converged Ethernet

ConnectX-2 EN 40G delivers the features needed

for a converged network with support for Data

Center Bridging (DCB). Fibre Channel frame

encapsulation compliant with T11 and hardware

ofoads simplies FCoE deployment. RDMA over

Converged Ethernet (RoCE) running over DCB

fabrics provides efcient low latency RDMA

services. The RoCE software stack maintains

existing and future compatibility with bandwidth

and latency sensitive applications. By maintaining

link-level interoperability with existing Ethernet

networks, IT managers can leverage existing data

center fabric management solutions.

I/O Virtualization

ConnectX-2 EN 40G supports hardware-based

I/O virtualization, providing dedicated adapter

resources and guaranteed isolation and protec-

tion for virtual machines (VM) within the server.

ConnectX-2 EN 40G gives data center managers

better server utilization and LAN and SAN unica-

tion while reducing costs, power, and complexity.

Quality of Service

Resource allocation per application or per VM

is provided by the advanced QoS supported by

ConnectX-2 EN 40G. Service levels for multiple

trafc types can be based on IETF DiffServ or IEEE

802.1p/Q, along with the Data Center Bridging

enhancements, allowing system administrators

to prioritize trafc by application, virtual machine,

or protocol. This powerful combination of QoS

and prioritization provides the ultimate ne-grain

control of trafc – ensuring that applications run

smoothly in today’s complex environment.

ConnectX

-2 EN 40G

40 Gigabit Ethernet Adapter with PCI Express 2.0

BENEFITS

– 40Gb/s connectivity for servers and

storage

– Industry-leading throughput and latency

performance

– I/O consolidation

– Virtualization acceleration

– High-performance networking and

storage access

– Software compatible with standard

TCP/UDP/IP and iSCSI stacks

KEY FEATURES

– Single 40GigE port

•

QSFP for copper or ber connections

– PCI Express (up to 5GT/s)

– RDMA over Converged Ethernet (RoCE)

– Data Center Bridging support

– T11 FCoE frame support

– TCP/UDP/IP stateless ofoad in

hardware

– Trafc steering across multiple cores

– Hardware-based I/O virtualization

– Intelligent interrupt coalescence

– Advanced Quality of Service

– Interoperable with 10GigE fabrics,

using QSA

– RoHS-R6

HIGHLIGHTS

PRODUCT BRIEF

SOFTWARE

MLNX-OS™ is a switch-based software tool that enables the

management and conguration of Mellanox SX6000 switch platforms

providing optimal performance for cluster computing, enterprise data

centers (EDC) and cloud computing.

The switch management capabilities ensures

the highest fabric performance while the chassis

management ensures the longest switch up time.

With MLNX-OS running on Mellanox’s switches,

IT managers will see a higher return on their

compute as well as infrastructure investment

through higher CPU productivity due to higher

network throughput and availability.

Switch Chassis Management

Mellanox advanced chassis management software

provides all the parameters and information IT

manager will need including: port status with

event and error logs, graphical CPU load display,

graphical fan speed over time display, power

supplies voltage alarms, graphical internal

temperature display with alarm notication and

more. These chassis management capabilities will

ensure low switch maintenance and high network

availability.

Subnet Management

MLNX-OS subnet management provides a

reliable and scalable management solutions

for cluster and data center fabrics. Its modular

design integrates the subnet manager (SM) with

advanced features simplifying cluster bring up and

node initialization through automatic discovery

and conguration. The performance monitors

measure the fabric characteristics to ensure the

highest effective throughput. For ease of use,

switch management is differentiated into basic,

advanced and expert modes where support for

various routing algorithms, QoS attributes and

fabric topologies are included.

World-Class Design

MLNX-OS software includes: CLI, WebUI, SNMP

and XML gateway interfaces. The XML Gateway

provides an XML request-response protocol that

can be used by end-user tools to get and set

management information on the appliance. The

service can be accessed over HTTP / HTTPS or

over SSH. The management also enables the user

to store all data into dened logs, dene e-mail

alerts and security capabilities such as: RADIUS,

TACACS+, AAA and LDAP.

Mellanox Advantage

Mellanox Technologies is a leading supplier of

end-to-end servers and storage connectivity solu-

tions to optimize data center performance and

efciency. Mellanox InniBand adapters, switches,

and software are powering Fortune 500 data

centers and the world’s most powerful supercom-

puters. The company offers innovative solutions

that address a wide range of markets including

HPC, enterprise, mega warehouse data centers,

cloud computing, Internet and Web 2.0.

MLNX-OS

Integrated Switch Management Solution

– Embedded Subnet Manager (SM)

– Reliable and scalable architecture

supporting up to 648 nodes

– Accelerated fabric design and

installation

– “out of the box” experience

– In-band and out-band support for

standalone or remote conguration

with secure access

– Hardware monitoring and alarms

– Performance monitoring

– Quality of Service based on trafc

type and service levels

– CLI, SNMP, WebUI and XML gateway

user interfaces

– E-mail alerts

HIGHLIGHTS

PRODUCT BRIEF

SWITCH SILICON

Mellanox continues its connectivity leadership by providing the highest

performing server and storage system interconnect solution for Enterprise

Data Centers, Cloud Computing, High-Performance Computing, and

embedded environments.

SwitchX, the fth generation switching IC from

Mellanox, further improves the company’s tradi-

tion of leading bandwidth, low-latency, and scal-

ability by integrating Virtual Protocol Interconnect

(VPI) technology - allowing InniBand, Ethernet

and Fibre Channel trafc to exist on a single

‘one-wire’ fabric. The industry-leading integra-

tion of 144 SerDes which are exible enough

to run 1 Gigabit to 14 Gigabit speeds per lane,

independent of network protocol, makes SwitchX

an obvious choice for OEMs that must address

end-user requirements for faster and more robust

applications. Reduced power, footprint and a fully

integrated PHY capable of connectivity across

PCBs, backplanes as well as passive and active

copper/bre cables allow interconnect fabrics

based on SwitchX to be utilized by network archi-

tects deploying leading, fabric-exible computing

and storage systems with the lowest TCO.

FDR

InniBand FDR technology moves from 8b10b

encoding to more efcient 64/66 encoding while

increasing the per lane signaling rate to 14Gb/s.

Using FDR 4x ports and Mellanox ConnectX

family of InniBand adapters, users can increase

end-to-end bandwidth up to 85% over previous

QDR solutions. Mellanox end-to-end systems can

also take advantage of the efciency of 64/66

encoding on 4x10Gb/s links using Mellanox

FDR-10.

SwitchX

™

Virtual Protocol Interconnect

Converged Switch-Silicon IC Device

36-port Switch-Router Device Provides Leading FDR 56Gb/s InfiniBand,

40 Gigabit Ethernet and Fibre Channel Throughput

– Industry-leading Cut Through Latency

– I/O Consolidation

•

IB and IP Routing

•

Ethernet and FC Gateways

– Virtualization Support

•

VEB, VEPA (+), Port Extender

– Data Center Bridging (DCB)

•

PFC, DCBX, ETS

– Low Cost Solution

•

Single-Chip Implementation

– Fully Integrated Phy

•

Backplane and cable support

•

1, 2 and 4 Lane

– Up to 4Tb/s switching capacity

– Flexible Port Congurations

•

Up to 36, 56Gig IB, 40GigE Ports

•

Up to 64 10GigE Ports

•

Up to 24 2/4/8Gig FC Ports

– Adaptive Routing

– Congestion Control

– Quality of Service 802.1p, DIFFSERV

– Switch Partitioning

– Multichip Support

•

All ports support stacking

•

Management across multiple devices

– Energy Efcient Ethernet

– IEEE 1588 Clock Synchronization

– Active Power Governor

HIGHLIGHTS

Management

Mellanox VPI™ I/O

Up to 36 Port FDR/40GigE - 64 Port 10/20GigE - 24 Port 2/4/8 Gig FC

SwitchX™

Non-blocking VPI™ Switch-Router-Gateway

Mellanox Integrated Flex SerDes

Active/Passive Copper - Fibre - PCB - Backplane

Adaptive

Routing

L2/L2+

Congestion

Control

QoS

Routers/

Gateways

Virtualization

Mellanox IRISC Processing Engine

Dual SGMII

PCIe

Figure 1: Mellanox SwitchX Architecture

InfiniBand Switch System Family

Highest Levels of Scalability,

Simplified Network Manageability,

Maximum System Productivity

Flexible I/O for the Dynamic Data Center

Mellanox 10 and 40 Gigabit Ethernet Converged Network Adapters

Uncompromising Integrity

Achieve full performance from your network equipment

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 13

PRODUCT BRIEF

SWITCH SYSTEM

The IS5030 switch system provides the highest-performing

fabric solutions in a 1RU form factor by delivering 2.88Tb/s of

non-blocking bandwidth to High-Performance Computing and

Enterprise Data Centers, with 100ns port-to-port latency.

Built with Mellanox’s 4th generation InniScale

InniBand switch device, the IS5030 provides up

to 40Gb/s full bidirectional bandwidth per port.

This stand-alone switch is an ideal choice for

top-of-rack leaf connectivity or for building small

to medium sized clusters. The IS5030 is designed

to carry converged LAN and SAN trafc with the

combination of assured bandwidth and granular

Quality of Service (QoS).

Sustained Network Performance

The IS5030 enables efcient computing with

features such as static routing, adaptive routing,

and congestion management. These features

ensure the maximum effective fabric bandwidth

by eliminating congestion hot spots. Whether

used for parallel computation or as a converged

fabric, the IS5000 family of switches provides the

industry’s best trafc-carrying capacity.

World-Class Design

The IS5030 is an elegant top of the rack design,

that are designed for performance, serviceability,

energy savings and high-availability. The design is

optimized for tting into industry-standard racks

and also for scale-out computing solutions from

industry leaders. The IS5030 supports reversible

airow making the design t into data centers

with different thermal designs. Redundant and

hot swappable power supplies and fans provide

high availability for both High-Performance

Computing and Enterprise Data Center applica-

tions. Status LEDs for fans and power supply units

are placed on both sides of the system for easy

status identication. Quick start guide, rack mount

kit, AC power cord and RJ45 to DB9 console

cables are provided as part of the standard

system providing new levels of out of the box

experience and performance.

Management Capabilities

The IS5030 comes with an onboard subnet

manager, enabling simple, out-of-the-box fabric

bring up for up to 108 nodes. FabricIT™ fabric

management running on the IS5030 enables fast

setup for cluster bring-up, fabric-wide bandwidth

and latency verication, and various diagnostic

tools to check node-node, node-switch connec-

tivity and cluster topology view. FabricIT™ chassis

management provides administrative tools to

manage the rmware, power supplies, fans, ports,

and other interfaces.

IS5030

36-port Non-blocking Managed 40Gb/s InfiniBand Switch System

KEY FEATURES

BENEFITS

– Industry-leading, switch platform in

performance, power, and density

– Designed for energy and cost savings

– Ultra low latency

– Granular QoS for Cluster, LAN and

SAN trafc

– Maximizes performance by removing

fabric congestions

– Fabric Management for cluster and

converged I/O applications

KEY FEATURES

– 36 QDR (40Gb/s) ports in a 1U switch

2.88Tb/s switching capacity

– Quality of Service enforcement

– Redundant power supplies and fan

drawers

MANAGEMENT

– Fast and efcient fabric bring-up

– Comprehensive chassis management

– Intuitive CLI and GUI for easy access

HIGHLIGHTS

IS5030

Documents and Marketing Collateral

PRODUCT BRIEF SPECIFICATIONS - FIRST PAGE

01 Key Feature Head p1

Bell Gothic Std - Bold

ALL CAPS,12 pt

Centered, White

02 Intro Body

Univers LT 47 CondensedLt

14 pt /18

Align Left, 85% Black

Logo Placement

X = 6.75”

Y = .5”

01 Folio

Bell Gothic Std - Bold

ALL CAPS, 14 pt

Centered, White

01 Folio2

Bell Gothic Std - Bold

ALL CAPS, 11 pt

Centered, White

01 Main Title

Univers LT 59 UltraCond

32 pt / 28

Align Left, Black

Dotted Rule below

02 Body Text

Univers LT 47 CondensedLt

10 pt / 12.75

Align Left, Black

02 Body Heading1

Univers LT 59 CondensedLt

Bold

10.5 pt / 12

.02” Space Before

Align Left, Color

05 Copyright

Univers LT 47

32 pt / 28

Align Left, Black

01 Main Title 2

Univers LT 45 Light

14 pt / 15

Align Left, Black

Product Image Area

Can overlap dotted line

01 Key Features/Apps

01 Key Features/Apps L2

01 Key Feature Head 2

Univers LT 47 CondensedLt

9.5 pt /11.5

Align Left, Black

.5”

.75”

@ 3.25”

2” from top

Dotted Line 3.25” from top

.1875”

100% PMS 282

Width = 2.25”

Highlights box started

at 6.75” and bleeds to

right edges

Main content area is 2.5” (2 columns starting at .5”

and extending to 6.5”). Gutter is .25”

Dotted Line

1 pt.

Black, dotted

.125” corners

All rounded corners shoud be .125” radius. All image boxes should be .5 pt, 100% black

01 Key Features/Apps

Univers LT 47 CondensedLt

9.5 pt /11.5

Align Left

Style Sheet Name in Template

Font

Font Size / Leading

Alignment

Item should be on Master page

CORPORATE STYLE GUIDE

14 | Mellanox Technologies | July 2011

Product Name

Univers LT 55 Bold

8.25 pt/9.5

SwitchX™

Virtual Protocol Interconnect

Converged Switch-Silicon IC Device

page 2

350 Oakmead Parkway, Suite 100, Sunnyvale, CA 94085

Tel: 408-970-3400 • Fax: 408-970-3403

www.mellanox.com

FabricIT is a trademark of Mellanox Technologies, Ltd. All other trademarks are property of their respective owners.

Virtual Protocol Interconnect (VPI)

SwitchX VPI devices enable industry standard

networking, clustering, storage, and manage-

ment protocols to seamlessly operate over a

single “one-wire” converged network. With

auto-sense capability, each SwitchX port can

identify and operate InniBand, Ethernet,

Data Center Bridging (DCB) or Fibre Channel

protocol. Combined with Mellanox’s ConnectX

family of VPI adapters, on-the-y fabric repur-

posing can be enabled for Cloud, Web2.0, EDC

and Embedded environments providing “future

proong” of fabrics indpendent of protocol.

Congurations

SwitchX allows OEMs to deliver:

– 36 Port 1U FDR IB switch

– 36 Port 40GigE or 64 Port 10GigE L2, L2+

and L3 switch

– 48 Port 10GigE to 12 Port 40GigE Top-of-

Rack switch

– Blade switches for converged fabrics

– (16 - 40GigE to servers, 12 - 10GigE to

LAN, 8 - 8G FC to SAN and 2 - 40GigE

stacking ports)

– Modular switch chassis up to 648 56G

IB/40GigE ports

Switch Product Development Platforms

The SwitchX Evaluation Board (EVB) and

Software Development Kit (SDK) are avail-

able to accelerate OEMs’ time to market

and for running benchmark tests. These rack

mountable systems are available with a mix

of QSFP and SFP+ connector for verifying

InniBand, 10GigE, 40GigE and Fiber Channel

functionality. In addition, SMA connectors are

available for SerDes characterization.

Mellanox Advantage

Mellanox is the leading supplier of industry

standard InniBand and Ethernet HCA, NIC

and switch silicon. Our products have been

deployed in clusters scaling to thousands-of

nodes and are being deployed end-to-end in

data centers and Top500 systems around the

world.

Ordering Part Number InniBand 4X Port Speed Typical Power

MT51236A0-FCCR-F SwitchX, 36 Port FDR Switch IC 72W

MT51336A0-FCCR-F SwitchX 36 Port VPI FDR/40GigE Switch IC 72W

MT51336A0-FCCR-Q SwitchX 36 Port VPI QDR/10GigE Switch IC 55W

INFINIBAND

– IBTA Specication 1.2.1 compliant

– 10, 20, 40 or 56Gb/s per 4x port

– Integrated SMA/GSA

– Hardware-based congestion control

– 256 to 4Kbyte MTU

– 9 virtual lanes: 8 data + 1 management

ENHANCED INFINIBAND

– IB to IB Routing

– Hardware-based adaptive routing

– Up to 8 subnets/switch partitions

– Fine grained end-to-end QoS

– Port mirroring

– Supports Jumbo frameup to 10KB

EThERNET

– 1, 10, 20 and 40Gb/s

– DCB (PFC, ETS, DCBX)

– FCoE

COmPaTIBILIITy

CPU

– PowerPC, IntelX86, AMDX86 and MIPS

PCI EXPRESS INTERFACE

– PCIe Base 2.0 compliant, 1.1 compatible

– 2.5GT/s or 5GT/s link rate x4

CONNECTIVITY

– Interoperates with InniBand, Ethernet and

Fiber channel adapters and switches

– Drives active/passive copper cables, ber

optics, PCB or backplanes

MANAGEMENT AND TOOLS

– Supports Mellanox UFM and IBTA compliant

Subnet Managers

– Diagnostic and debug tools

I/O SPECIFICaTIONS

– 36 4x SDR/DDR/QDR/FDR InniBand ports,

36 40GigE ports, 64 10/20GigE ports, 24

2/4/8Gig FC ports or a combination of port

types

– PCI Express 2.0 x4 5GT/s (1.1 compatible)

– SPI Flash interface, I

– IEEE 1149.1 boundary-scan JTAG

– Link status LED indicators

– General purpose I/O

– 45 x 45mm FCBGA

HaRDWaRE

01 Key Feature Head p1

Bell Gothic Std - Bold

ALL CAPS,12 pt

Centered, White

02 Body Text

Univers LT 47 CondensedLt

10 pt / 12.75

Align Left, Black

02 Body Heading1

Univers LT 59 CondensedLt

Bold

10.5 pt / 12

.02” Space Before

Align Left, Color

Corporate Address

Univers LT 57 Condensed

9 pt/12

.875”

1.5”

.5”

Dotted Line

1 pt.

Black, dotted

.125” corners

Documents and Marketing Collateral

PRODUCT BRIEF SPECIFICATIONS - LAST PAGE

All rounded corners should be .125” radius. All image boxes should be .5 pt, 100% black

01 Key Features/Apps

Univers LT 47 CondensedLt

9.5 pt /11.5

Align Left

Style Sheet Name in Template

Font

Font Size / Leading

Alignment

Item should be on Master page

01 Second Title

Univers LT 55

8.25 pt/9.5

03 Page No.

Univers LT 47 CondensedLt

8 pt/9.5

01 Key Features/Apps

01 Key Features/Apps L2

01 Key Feature Head 2

01 Key Feature Head 2b

Univers LT 47 CondensedLt

Univers LT 47 CondensedLt Bold

9.5 pt /11.5

Align Left, Black

04 Table Heading

Univers LT 47 CondensedLt Bold

8.5 pt /9

Centered, Black

03 Copyright

Univers LT 47 CondensedLt

6 pt /7

Align Left, Black

03 Doc No.

Univers LT 47 CondensedLt

5 pt /5

Align Left, 80% Black

04 Table Text - FL

04 Table Text - Ctr

Univers LT 47 CondensedLt Bold

8.25 pt /9.25

80% Black

Table Rules

.5 pt

60%

No rule on left and

right outside columns

XXXXPB Rev 1.0

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 15

1$."(9

!"#$%#&'()*

!"#$%&'#&()*'+!,)-.(/0!123!45!3%!6+(!7-#.8/

8(>&16*/.4-(9:;;7(!"##$%&'(<"04%&#&./")7(=##(*/.4-)(*")"*+",7

!"*-$-"!!&-$&-!,!%&-!,!$-!,!"%&-!,!-!,!-!,!-+-$&'-$"&""-!&$"!!&-!-"&$-$-$%&$-&$ $%-"-!"*-!""%-&-

$-%--&$ $-"-!"*-!""%-&--"&$-&$ $%-$-#$"#$&+-"-&$-$%#&(-")!$%

!"#$%&'()&*$+&,'-&./$0123)$4##/$0155.6&7)/$89$:;#<"





!"#$%&'()*%7,%82392.$%!1:;".*

Fewer I/O adapters, cables and switches – 80% TCO

savings

!>)*#<2]"'#2)'5)7%#("<&270#F2*#1*)78#^)82$)4#[4>%'7%4#

X^)8[Y#27=#Z)$'%#\>277%5#XZ\Y#.DE#2=2&4%'#(2'=*#27=#

*F)4(>%*#)7#4>%#=242(%74%'#F>)(>#("<&')*%=#"/#,?#

_H#O52=%#[7(5"*1'%*#F)4>#SP#_H#O52=%#@%'3%'*-#[2(>#

(>2**)*#)7(51=%=#J#Z\#&"'4*#27=#,`#^)8[#&"'4*-#!>%0#

F%'%#("77%(4%=#1*)78#J`#O'"(2=%#W\@Q$2*%=#&"'4*#

27=#?,P#\)*("#\24250*4#PLMSQ$2*%=#&"'4*#'%*&%(4)3%50-#

_H#a)'4125#\"77%(4#<"=15%*#27=#5)(%7*%*#F%'%#1*%=#4"#

=%5)3%'#3)'4125)6%=#.DE#*%'3)(%*-#V>%7#4>%#Z\#27=#^)8[#

&"'4*#F%'%#'%&52(%=#$0#:%5527";Q$2*%=#17)/)%=#.DE#

&"'4*9#%2(>#_H#O52=%#(>2**)*#("15=#7"F#="#F)4>#]1*4#J#

:%5527";Q$2*%=#.7/)7)O27=#.DE#&"'4*#F)4>"14#'%K1)')78#

270#(>278%#4"#1*%'#2&&5)(24)"7*#27=#%;&%')%7(%-#B"F9#

"750#J#O'"(2=%#W\@Q$2*%=#&"'4*#27=#,P#\)*("#\24250*4#

PLMSQ$2*%=#&"'4*#2'%#7%%=%=-#!>)*#'%*154%=#)7#(2&)425#

%;&%7=)41'%#*23)78*#"/#T,-?:#27=#27"4>%'#T?MM9MMM#)7#

"&%'24)"725#%;&%7*%#*23)78*#/'"<#/5""'#*&2(%#27=#&"F%'#

'%=1(4)"7*#X2<"'4)6%=#"3%'#G#0%2'*Y-

!"#$%&'()*%<,%=$##%'>$%>"9?%'>$%.(:@$3%1?%#$34$3#%

)$924$3%'>$%#":$%'3".#"5'21.%3"'$%A%B7CD-%E!F%

#"42./#

63% more transactions/second with half the servers –

$2.6M TCO savings

!>)*#3%'0#52'8%#&2(I28%#*>)&&)78#("<&270#2&&5)%=#

:%5527";#.7/)7)O27=Q$2*%=#*%'3%'#4"#*%'3%'#)74%'("77%(4#

&'"=1(4*#X2=2&4%'*#27=#*F)4(>%*Y#/"'#&'"(%**)78#E'2(5%#

bA\#=242$2*%Q$2*%=#)73")(%#&'"(%**)78#F"'I5"2=*-#O0#

=")78#*"9#4>%#("<&270#F2*#2$5%#4"#2(>)%3%#PGN#<"'%#

)73")(%#&'"(%**)78#4'27*2(4)"7*#&%'#*%("7=9#255"F)78#

)4#4"#*%'3)(%#)4*#(1*4"<%'*#/2*4%'#27=#$%44%'-#!>%#!\E#

*23)78*#("<&')*)78#"/#*%'3%'9#.DE9#*"/4F2'%#5)(%7*%#27=#

<2)74%727(%#%;&%7*%*#"3%'#2#J#0%2'#2<"'4)624)"7#&%')"=#

2<"174%=#4"#2#F>"&&)78#T?-P:-

!"#$%&'()*%G,%H2.".52"9%&$3425$#%!1:;".*

82% more transactions/second at 70% lower costs and 3

times less power

b2&)=#8'"F4>#"/#=242#"7#(2&)425#<2'I%4*#27=#)7('%2*)78#

1*%#"/#214"<24%=#4'27*2(4)"7*#*1(>#2*#258"')4><)(#

4'2=)78#)*#)7('%2*)78#4>%#&'%**1'%#"7#/)727()25#)7*4)414)"7*#

F"'5=F)=%#4"#1&8'2=%#4>%)'#.!#)7/'2*4'1(41'%*#4"#*420#

("<&%4)4)3%-#A7#%;2<&5%#"/#*1(>#27#2&&5)(24)"7#1*%=#

$0#4>%#52'8%#/)727()25#)7*4)414)"7#)7#4>)*#(2*%#*41=0#)*#4>%#

b%14%'*#:2'I%4#W242#@0*4%<#Xb:W@Y#F>)(>#*%'3%*#

2*#2#&524/"'<#/"'#<2'I%4#=242#=)*4')$14)"7-#.7#%2(>#"/#

4>%#(')4)(25#<%4')(*#4>24#>23%#2#=)'%(4#$%2')78#"7#4>%#

&%'/"'<27(%#"/#4>%#/)727()25#)7*4)414)"79#:%5527";#

&'"=1(4*#=%5)3%'#%;(%55%74#3251%#c#`?N#>)8>%'#1&=24%*#

&%'#*%("7=9#P?N#5"F%'#<%27#524%7(0#24#UMN#5"F%'#

("*4*#27=#G#4)<%*#5"F%'#&"F%'#("7*1<&4)"7-#

!"#$%&'()*%I,%6324"'$%!91()%8;;92".5$%63142)$3

Better VM replication and DB scaling with 75% lower

hardware acquisitions costs

!>)*#(2*%#*41=0#)73"53%*#2#&')324%#(5"1=#2&&5)27(%#

&'"3)=%'#4>24#=%5)3%'*#4>%#*2<%#$%7%/)4*#"/#@"/4F2'%#

2*#2#@%'3)(%#X@22@Y#F>)5%#255"F)78#(5"1=#(1*4"<%'*#

4"#<2)742)7#("<&5)27(%#27=#=242#*%(1')40#)7#4>%)'#

"F7#=242(%74%'*-#!>)*#3%'4)(25#*"514)"7#)7(51=%*#

)74%8'24%=#*"514)"7*#/"'#.7/'2*4'1(41'%#2*#2#@%'3)(%#

X.22@Y9#H524/"'<#2*#2#@%'3)(%#XH22@Y#27=#@"/4F2'%#2*#2#

@%'3)(%#X@22@Y9#F)4>#:%5527";#)74%'("77%(4#=%5)3%')78#

27#%//)()%749#*(252$5%#27=#>)8>#&%'/"'<)78#)74%'("77%(4#

)7#4>%#.22@#520%'-#.7#40&)(25#=%&5"0<%74*#*1(>#2*#F)4>#

2#52'8%#d2&27%*%#("785"<%'24%9#7%4F"'I#>2'=F2'%#

2(K1)*)4)"7#("*4*#F%'%#'%=1(%=#$0#1&#4"#LMN#27=#

7%4F"'I#<2728%<%74#("*4*#$0#1&#4"#GMN-#A4#4>%#

*2<%#4)<%9#:%5527";#&'"=1(4*#='2<24)(2550#'%=1(%=#

4>%#&%'/"'<27(%#)<&2(4#"/#'%25Q4)<%#3)'4125#<2(>)7%#

'%&5)(24)"7#/"'#/2)5#"3%'#&1'&"*%*9#27=#4>'"18>#1*%#

"/#17)/)%=#<%<"'0#*&2(%#2('"**#*%'3%'*9#=242$2*%#

2&&5)(24)"7*#F%'%#*(25%=#1&#%//)()%7450#4>'"18>#1*%#"/#

/%F%'#*%'3%'*-#E3%'2559#>2'=F2'%#2(K1)*)4)"7#("*4*#F%'%#

'%=1(%=#$0#ULN-

CASE STUDY

Mellanox server and storage interconnect products deliver down

to earth values that relieve the most critical pain points in the data

center today.

To be precise:

1. If you are virtualizing your servers, Mellanox

simply delivers you more virtual machines

per server

2. If you are connecting your servers to

the SAN and LAN, Mellanox reduces

the number of I/O adapters, cables and

switches

3. If you are running data warehousing, OLTP,

or financial services applications, Mellanox

delivers more transactions per second

using fewer servers, helping you make

faster and better decisions

4. If you are a cloud appliance or services

provider, Mellanox reduces equipment

acquisition costs dramatically while

improving your ability to exceed your SLAs

with customers by a handsome margin

5. Mellanox delivers all of the above benefits

while reducing your power consumption

for IO by 30% to 50%

These are a few examples of how Mellanox

is helping change the face of the data center

with high performance, scalable and efficient

server and storage connectivity products.

Below are five proof points relating how

fortune 500 customers are availing these

benefits to save money, build more efficient

and greener datacenters thereby improving

their competitive edge through better service

delivery to their end users and customers.

Case Study 1: Managed Hosting Services

Provider

60% more VMs per server --> 60%+ ROI beneﬁts

The customer is a large managed hosting

services provider servicing web based travel

related transactions for their customers. The

volume of transactions per day is greater than

that of Amazon.com, requiring the customer

to manage its data center efficiently and scale

in a cost effective way. One of its datacenter

components serviced customers through the

use of applications in 256 virtual machines,

which prior to using Mellanox-based

solutions, required the use of 4 racks of 4U

servers, 16 edge switches, and 192 I/O cards.

The capital cost to build out that infrastructure

was $744,000. With Mellanox-based

products, without requiring any changes to

their applications or affecting services to its

customers, the new component build-out

required only $347,000, saving $397,000 in

capital expenses. The cost savings came

from the use of 1 rack of 1U servers, 2 I/O

Directors and 32 I/O cards that could still

support the 256 virtual machines. Operating

expenses came down through reduced floor

space and power requirements, bringing total

savings to more than $500,000 (amortized

over 3 years).

Down-to-Earth Values, ROI Up in the Clouds

Featuring Mellanox Interconnect Products for the Enterprise Data Center

OVERVIEW

Mellanox is helping change

the face of the data center

with high performance,

scalable and efficient server

and storage connectivity

products.

Documents and Marketing Collateral

CASE STUDIES

01 Main Title

Univers LT 59 UltraCondensed

Regular

28 pt/30

01 Main Title 2

Univers LT 45 Light

14 pt/16 pt

Tracking: -10 pt

03 Callout

Univers LT 45 Light

Italic

10 pt/16 pt

PMS 280

02 Intro

Univers LT 47 CondensedLt

Regular

13 pt/17

PMS 280

02 Body Copy

Univers LT 47 CondensedLt

9.5 pt/12

02 Body Heading 1

Univers LT 47 CondensedLt,

Bold

11.5 pt/13

02 Body Heading 2

Univers LT 47 CondensedLt,

Bold

10 pt/11.5

See Product Brief Specs

Frame

1 pt

75% Black

Japanese Dots (InDesign)

Corners: .125” radius

.5”

2.125”

.875”

6.0” - 2 column format, .1875” gutter

Width = 2”

.75”

2”

.5”

1.25”

.5”

INDUSTRY ICON

01 Key Feature Head p1

Bell Gothic Std - Bold

ALL CAPS,12 pt

Centered, White

CORPORATE STYLE GUIDE

16 | Mellanox Technologies | July 2011

DATASHEET

CABLES

Mellanox’s 40GigE Passive copper cables provide robust connection to switches and network adapters

complying with 40GBase-CR4 specications. The 40GigE passive copper cables are compliant with the

IEEE 802.3ba and, SFF-8436 specications and provide connectivity between devices using QSFP ports.

40GigE passive copper cables ll the need for short, cost-effective connectivity in the data center. Mella-

nox’s high-quality passive copper cable solutions provide a power-efcient replacement for active power

connectivity such as ber optic cables.

Optimizing systems to operate with Mellanox’s 40GigE passive copper cables signicantly reduces poer

consumption and EMI emission, eliminating the use of EDC hosts.

Table 1 - Cable Specications and Ordering Information

Maximum

Reach [m]

Gauge Max Cable

Diameter [mm]

Min Bend

Radius [mm]

Assembly

Min Ben

Radius [mm]

Weight

[Grams/meter]

Part Number

1 30 6.2 30 61 55 MC2210130-001

2 30 6.2 30 61 55 MC2210130-002

3 30 6.2 30 61 55 MC2210130-003

4 28 7.8 40 71 79 MC2210130-004

5 28 7.8 40 71 79 MC2210130-005

1. See gure 1 for bend radius description

2. Weight is per raw cable meter, excluding connectors

Table 2 - Absolute Maximum Ratings

Parameter Parameter Range

Operating Case Temperature 0°C to 70°C

Storage Ambient Temperature -10°C to 75°C

Table 3 - Operating Conditions

Parameter Minimum Typical Maximum Units

Data Rate per Lane 1 10 Gb/s

Input Voltage -0.3 3.6 V

Humidity (RH) 5 85 %

40Gig Ethernet, Passive Copper Cables

– Compliant with SFF-8436

– Up to 5m 40GigE data rates

– Ultra low crosstalk for improved

performance

– Low insertion loss

– BER better than 10-15

– Serial numbers printed on each end

– Tested in an end-to-end 40GigE system

– RoHS 6 Compliant

HIGHLIGHTS

Documents and Marketing Collateral

DATASHEETS

Image Rule

Size: .5 pt

Color: Black

01 Main Title

Univers LT 59 UltraCondensed

32 pt/34

Tracking: -10

Aligned Left

04 Table Title

Univers LT 55

10 pt/12

Aligned Left

03 Figure Caption

Univers LT 47 CondensedLt, Italic

9 pt/10.75

80% Black

Aligned Left

03 Figure Caption

Font Family: Univers LT 47 CondensedLt

Italic

9 pt/10.8

Figure X. Bold Italic

02 Body Text

Univers LT 47 CondensedLt

10 pt /12.75

Tracking: -3 pt

!"#"$%&&#



>)0'%"

!"#$%&'()&*$+&,'-&./$0123)$4##/$0155.6&7)/$89$:;#<"





!%G+>S510/6%"#$$3%&'(()*+,%-'./*+(+01'23%4((%510/62%5'2'57'83

="55"64;'E6FG5%"4$G6

$!5!'')*25--%0!5*++!,5%#55'!-5%)'/ !553!,5

(1L16'8%/)58A)5'%A)55)*6S\%A/1./%.+7'52%5'>)15%+5%5'>().'L'*63

:88$4$G6"2'E6FG5%"4$G6

*,5(*,!5%)"*,(.%*)5*/.5!'')*255'!5-*'/.%*)-5+'!-!5

.+*6).6%S+@5%&'(()*+,%-'./*+(+01'2%2)('2%5'>5'2'*6)617'%+5%71216]%

/66>]OOAAA3L'(()*+,3.+L^[email protected]^_G)?('2^_G)?('2%K7'571'A

5!05

!"#$%&'()'

!"#$%&'*)'

See Product Brief Specs

2.25”

5.75” - 1 column format

See Product Briefs and

Case Study Specs

See Product Briefs

Specs

Table Specs

See Product Briefs

Table Rules

.5 pt, 60%

No rule on left and right outside columns

CORPORATE STYLE GUIDE

Mellanox Technologies | July 2011 | 17

– The world’s fastest interconnect, supporting

up to 40Gb/s per adapter

– Latency as low as 1 microsecond

– Full CPU ofoad with the exibility of RDMA

capabilities to reduce traditional network

protocol processing from the CPU and increase

the processor efciency.

– I/O Capex reduction – one 40Gb/s Mellanox

adapter carries more trafc with higher

reliability than four 10 Gigabit Ethernet

adapters.

KEY ADVANTAGES

SOLUTION BRIEF

Computer Aided Engineering (CAE) is used to

help manufacturers bring products to market

faster while maintaining a high-level of quality. The

faster companies can conduct tests and perform

product analysis, the bigger the benefits of using

CAE. Advances in software and server hardware

have set the stage for faster results, but manufac-

turers should not overlook a major performance-

robbing bottleneck: the server interconnect. To

gain dramatic improvements in CAE performance,

manufacturing firms are turning to Mellanox’s

InfiniBand-based solutions to speed the move-

ment of data between clustered servers.

The Manufacturer’s Challenge

Bringing products to market fast while meeting

quality requirements and adhering to safety stan-

dards has become a daunting challenge to manu-

facturers. To remain competitive, manufacturers

must deliver products as fast as possible. But if

quality suffers, customers won’t return. If safety

levels decline, significant recalls, lawsuits or

harmful publicity could ensue.

This is why manufacturing companies rely so

heavily on Computer Aided Engineering (CAE),

which helps simulate production and product

performance ahead of time. CAE allows problems

to be corrected before products reach the produc-

tion stage and end up in the hands of customers.

The challenge that has emerged today is how to

run commercially-available CAE software faster

and with more accuracy. Many software vendors

offer capable products, but bottlenecks commonly

occur in the hardware that runs the simulations

and analyzes the production processes.

Without sufficient computing power, these tests

sometimes take days and weeks to run. And be-

cause the tests take longer to run, product-devel-

opment teams are often forced to run fewer tests

in order to meet tight timeframes and to remain

competitive. This can lead to inaccurate testing

that often goes undetected until the products

hit the assembly line—when the cost to make

changes to the design of the products grows

exponentially.

Solving this application run-time challenge will

allow manufacturers to analyze their processes

in the fastest time possible and conduct more

granular testing. This in turn will allow for quicker

and more efficient production. A key element to

running tests more rapidly is choosing the right

hardware to run the CAE applications.

Today’s Solution

Large symmetric multi-processing machines

(SMPs) used to be the answer for generating

compute power in the data center. However,

these proprietary, expensive systems gave way to

cluster and grid architectures consisting of low-

cost commodity elements that offer comparable

performance.

Some companies try to scale CAE clusters

by adding more servers or moving to servers

with multiple cores. This approach can work for

smaller, simpler simulations, but the more com-

plex the analysis, the more likely the need to run

simulations across multiple servers where latency

is a major factor in determining performance.

The answer to speeding analysis and maximizing

return on CAE investments is not simply buying

more or bigger servers, but rather eliminating

bottlenecks to performance by employing the use

of a high performance interconnect.

Because of the ready-availability of Ethernet,

many of today’s clusters are built with Ethernet

as the interconnect. While Gigabit Ethernet-based

clustering is cheaper than SMP-based architec-

tures, it can be very inefficient. For applications

that rely on bandwidth or memory sharing, the ef-

ficiency (percentage of a server-CPU dedicated to

communications overhead) can be a concern.

Today, many manufacturers run CAE systems that

take advantage of InfiniBand-based high-speed

Manufacturing

Bringing New Levels of Performance to CAE Applications

Solution Partner

Documents and Marketing Collateral

SOLUTION BRIEFS

Sidebar Space

Sidebar can be used for

logos or product images

01 Key Features p1

Bell Gothic Std, Bold

10pt

CAPS

White

Aligned Left

10% Black Box

Inset .08 pt

Header

.125” top corner radius

Inset .625 pt

01 Key Features/Apps

Univers LT 47 CondensedLT

9 pt/11

Aligned Left

.875”

1.125”

2.125”

See Product Briefs and

Case Study Specs

page 2

SOLUTION BRIEF

interconnect support incorporated into CAE software. Mel-

lanox InfiniBand interconnects eliminate I/O bottlenecks

allowing applications to run faster and more efficiently.

02040607010 30 50

# Cores

46.59

75% Efficiency

33.00

51% Efficiency

Parallel Speedup

Linear Scaling

GbE

InﬁniBand

Figure 1. Mellanox InniBand-based solution improves performance by

50%

A Better Way

Mellanox InfiniBand solutions help CAE applications run

faster. Mellanox offers high-performance (10, 20 and 40

Gbps), low-latency (< 2 microseconds) interconnect solu-

tions for CAE applications. Benchmark testing has found

that Mellanox interconnect solutions reduce CAE-runtime

by as much as 50 percent.

In addition to offering InfiniBand switch-technology, Mel-

lanox works directly with CAE software-vendors to create

the most efficient, fastest, and lowest-latency solutions in

the industry. By combining the leading CAE software with

Mellanox InfiniBand-based solutions, manufacturing orga-

nizations can now analyze products faster and more effi-

ciently to gain a clear competitive advantage.

As today’s price and performance leader in the industry,

Mellanox builds its solutions using standards-based In-

finiBand technology. InfiniBand is an industry-standard

interconnect for high-performance computing (HPC) and

enterprise applications. The combination of high band-

width, low latency and scalability makes InfiniBand the

interconnect-of-choice to power many of the world’s

largest and fastest computer systems and commercial

data centers. Mellanox solutions support most major

server vendors, operating systems, storage solutions and

chip manufacturers.

1 Gb

Ethernet

10 Gb

Ethernet

Myrinet InniBand

Bandwidth 1 Gb/sec 10Gb/sec 2.5 Gb/sec 10, 20 & 40

Gb/sec

Latency ~10 us 2.5 - 5.5 us <2 us

Average Efciency 53% No Entries 68% 74%

Price Per Gig/Port ~$350.00 >~$700.00 ~$225.00 <$100.00

Table 1. Price/performance advantages for InniBand

Building CAE Clusters

Mellanox offers complete end-to-end server interconnect

solutions for speeding CAE applications. The two major ele-

ments of the solution include:

• High-speed, low latency InfiniBand switches

• Fast storage connectivity

High-performance InniBand Switches

Mellanox’s InfiniBand-based solutions deliver high perfor-

mance and scalability to compute clusters. Mellanox offers

a complete portfolio of products including a scalable line of

InfiniBand switches, high performance I/O gateways (for

seamless connectivity to Ethernet and Fibre Channel net-

works) and fabric management software. Mellanox solu-

tions use the Open Fabric Alliance’s OFED drivers and the

OpenMPI (Message Passing Interface) libraries to optimize

application performance for both MPI-based and socket-

based applications.

For small-to-medium sized clusters, Mellanox offers the

Mellanox Grid Director

™

9024. It is a 1U device with

twenty-four 10 Gbps (SDR) or 20 Gbps (DDR) InfiniBand

ports. The switch is a high performance, low latency, fully

non-blocking edge or leaf-switch with a throughput of 480

Gbps.

Figure 2. Mellanox Grid Director 9024 for small-to-medium sized

clusters ranging from 16 to 24 nodes.

It is well-suited for small InfiniBand fabrics with up to 24

nodes because it includes all of the necessary manage-

ment capabilities to function as a stand-alone switch. The

Grid Director 9024 is internally managed and offers com-

prehensive device and fabric management capabilities.

Designed for high-availability (high MBTF) and easy mainte-

nance, the switch is simple to install and features straight-

forward initialization. The solution is scalable as additional

switches can be added to support additional nodes.

For larger clusters ranging from 25-96 compute nodes,

Mellanox offers the Grid Director

™

2004 multi-service

switch — the industry’s highest performing multi-service

switch for medium-to-large clusters and grids. The switch

enables high performance non-blocking configurations and

features an enterprise-level, high availability design. The

Grid Director 2004 supports up to 96 InfiniBand 4X ports

(20 Gbps) and is scalable through the use of additional, hot-

swappable modules. The Grid Director 2004 also features

10 GbE and Fibre Channel ports so the solution can provide

high-performance, integrated SAN and LAN connectivity.

Some Solution Briefs contain a Key Advantages box.

This element can be found in the InDesign library le.

CORPORATE STYLE GUIDE

18 | Mellanox Technologies | July 2011

Documents and Marketing Collateral

REFERENCE GUIDES

REFERENCE GUIDE

ConnectX

-2 Adapter Cards

Why Mellanox?

– Leading I/O consolidation for servers and storage!

– 40Gb/s end-to-end solutions

(HCAs, switches, cables)

– Proven scalability and reliability

– Field proven – 1000‘s of clusters in production, over 5M ports

deployed!

– Advanced features for highest application productivity

– High performance computing and enterprise and cloud data

centers, virtualized and unied systems

– 40Gb/s InniBand and 10 Gigabit Ethernet in the same device!

– World-class Support

– Best-in-class production testing and product quality

Why 40Gb/s InniBand?

Enables the highest performance and lowest latency

– Proven scalability for 10s of thousands of nodes

– Maximum return on investment

Highest Efciency / Maintains balanced system ensuring

highest productivity

–

No articial bottlenecks, performance match for PCIe Gen2

–

Proven to fulll multi-process networking requirements

–

Guaranteeing no performance degradation

Performance driven architecture

–

MPI latency 1us, 6.6GB/s with 40Gb/s InniBand (bi-directional)

–

MPI message rate of >40 Million/sec

Superior application performance

–

From 30% to over 100% HPC applications performance increase

–

Doubles the storage throughput, reducing backup time in half

InniBand Market Applications

InniBand is increasingly becomes an interconnect of choice in not

just high performance computing environments, but also in main-

stream enterprise grids, data center virtualization solutions, storage,

and embedded environments. The low latency and high performance

of InniBand coupled with the economic benets of its consolida-

tion and virtualization capabilities provides end-customers the ideal

combination as they build out their applications.

Data Center

Today’s data centers need an agile infrastructure that incorporates

ongoing improvements in computer, storage, networking, and

application technologies, and empowers IT to support changing

business processes. InniBand fabric solutions enable IT organiza-

tions to turn computing and storage resources from monolithic

systems to service-centric shared pool of resources consisting of

standardized components that can be dynamically provisioned and

accessed through an intelligent network.

High Performance Computing

With InniBand’s proven scalability and efciency, small and

large clusters easily scale up to thousands of nodes. With 40Gb/s

node-to-node and 120Gb/s switch-to-switch solutions available,

and a roadmap to 240Gb/s, InniBand has proven its performance

in personal supercomputing, workgroup, departmental, divisional,

and enterprise supercomputers solutions.

Storage

InniBand’s high bandwidth, low latency, dedicated I/O channels,

QoS and RDMA features can lower capital expenses and operating

costs making it the right choice for storage.

ConnectX-2 adapter cards with Virtual Protocol Interconnect (VPI) provide the highest performing and

most exible interconnect solution for Enterprise Data Centers, High-Performance Computing, and

Embedded environments. ConnectX-2 with VPI also simplies network deployment by consolidating

cables and enhancing performance in virtualized server environments. In addition, ConnectX-2 supports

CORE-Direct application acceleration, Virtual-IQ technology for vNIC and vHBA support, and congestion

management to maximize the network efciency making it ideal for HPC or converged data centers

operating a wide range of applications.

Key Features

– 1us MPI latency

– 10, 20, or 40Gb/s InniBand ports

– PCI Express 2.0 (up to 5GT/s)

– CPU ofoad of transport operations

– End-to-End QoS & congestion control

– Hardware-based I/O virtualization

– TCP/UDP/IP stateless ofoad

Key Advantages

– High-performance networking and storage

access

– Guaranteed bandwidth & low-latency

services

– Reliable transport

– End-to-End storage integrity

– I/O consolidation

– Virtualization acceleration

– Scales to tens-of-thousands of nodes

Mellanox 20 and 40Gb/s InniBand switches deliver the highest performance

and density with a complete fabric management solution to enable compute

clusters and converged data centers to operate at any scale while reducing

operational costs and infrastructure complexity. IS5000’s scalable switch

building blocks from 36 to 648 ports in a single enclosure gives IT managers the

exibility to build networks up to 10’s of thousands of nodes.

Key Features

– 51.8Tb/s switching capacity

– 100ns to 300ns switching latency

– Hardware-based routing

– Congestion control

– Quality of Service enforcement

– Up to 6 separate subnets

– Temperature sensors and voltage

monitors

Key Advantages

– High-performance fabric for

parallel computation or I/O

convergence

– Wirespeed InniBand switch

platform up to 40Gb/s per port

– High-bandwidth, low-latency

fabric for compute-intensive

applications

Delivers the industry’s lowest 1.3 us latency performance with performance-leading 10 and 40 Gigabit

Ethernet connectivity with stateless ofoads for converged fabrics in Enterprise Data Centers, High-

Performance Computing, and Embedded environments. Clustered databases, web infrastructure, and IP

video servers are just a few example applications that will achieve signicant throughput and latency

improvements resulting in faster access, real-time response and increased number of users per server.

ConnectX-2 EN improves network performance by increasing available bandwidth to the CPU and

providing enhanced performance, especially in virtualized server environments.

Key Features

– Industry-leading throughput and latency performance

– Enabling I/O consolidation by supporting TCP/IP, FC over Ethernet and RDMA over Ethernet

transport protocols on a single adapter

– Improved productivity and efciency by delivering VM scaling and superior server utilization

– Supports industry-standard SR-IO Virtualization technology and delivers VM pro tection and

granular levels of I/O services to applications

– High-availability and high-performance for data center networking

– Software compatible with standard TCP/UDP/IP and iSCSI stacks

– High level silicon integration and no external memory design provides low power, low cost and

high reliability

– High-bandwidth, low-latency fabric for compute-intensive applications

InniBand Switches

ConnectX EN Ethernet Network Interface Cards (NIC)

I",570,/I#,;5@,BI&>4=0I

&>88B?,60II

'06IIGI,AI

www.mellanox.com

REFERENCE GUIDE

HI9:B;423=II 066,89AI'0.38969240<I66I;423=<I;0<0;?0/

066,89AI;4/20*I9880.=*I"%4;0.=I8D846,<=I8D84;4/20I8D849<=I8D84%&I8D84&.,60I8D84#I#3B*I

)4;=>,6I#;9=9.96I8=0;.9880.=I,8/I)96=,4;0I,;0I;024<=0;0/I=;,/07,;5<I91I 066,89AI'0.38969240<I=/I,-;4.'I4<I,I=;,/07,;5I

91I 066,89AI'0.38969240<I=/I66I9=30;I=;,/07,;5<I,;0I:;9:0;=BI91I=304;I;0<:0.=4?0I9@80;<

 *$08 , &$/8-,1 "1

3>.5I'B->;

DirectorofSales

'06I

7,46I.3>.57066,89A.97

20(,$008$3$*-.+$,18-,1 "1

SteveWilliams

" I><480<<I0?069:708=I 82;

066I

7,46I<=0?07066,89A.97I

($*#8 /)$1(,&8-,1 "1

DarrinChen

&;I4;0.=9;I" II3,8806I406/I 5=2

066I

7,46I/,;;487066,89A.97

2..-/1$"',(" *8-,1 "1

GlennChurch

406/I::64.,=498<I82

066I

7,46I260887066,89A.97

4(1"'8601$+08/-#2"10 $** ,-58 /18 MFG

/#$/ !*$8

-/18 , &$#8884(1"'82,#*$

I%,.5I8D84&.,60

I)I%I8D84,8/I&@4=.3II$&#I:9;=<II#9@0;I&>::6BI

,8,20/I##*%I#&(I<4/0I=9I.9880.=9;I<4/0I,4;E9@I&=,8/,;/I/0:=3I%,46I

Kit.IncludesChassisManagement

MIS5030Q1SFC +

-/18 , &$#8884(1"'82,#*$88(58 ")

8D84&.,60

I)I$%I8D84,8/I&@4=.3II$&#I:9;=<II#9@0;I&>::6BI ,8,20/I

##*%I9880.=9;I<4/0I=9I#&(I<4/0I,4;E9@I&39;=I/0:=3I%,46I4=I8.6>/0<I

ChassisManagement

&$% +

-/18884(1"'82,#*$

8D84&.,60

I)I'-<I$%I8D84,8/I3,<<4<I&@4=.3I<>::9;=482I>:I=9II:9;=<I

8.6>/0<II&:480I6,/0<I ,8,20708=I 9/>60I##*II:9@0;I<>::640<I

andinstallationrails.IncludesChassisManagement;UpgradeabletoFabricIT

8=0;:;4<0I,-;4.I ,8,20;I@4=3I:>;.3,<0I91I,-;4.' I4.08<0

MIS5100Q3DNC +

-/18884(1"'82,#*$

8D84&.,60

I)I'-<I$%I8D84,8/I3,<<4<I&@4=.3I<>::9;=482I>:I=9II

:9;=<I8.6>/0<II&:480I6,/0<I ,8,20708=I 9/>60I##*II:9@0;I<>:

:640<I,8/I48<=,6,=498I;,46<I(:2;,/0,-60I=9I,-;4.'I8=0;:;4<0I,-;4.I ,8,20;I@4=3I

:>;.3,<0I91I,-;4.' I4.08<0

&$! +

-/18884(1"'82,#*$

8D84&.,60

I)I'-<I$%I8D84,8/I3,<<4<I&@4=.3I<>::9;=482I>:I=9II

:9;=<I8.6>/0<II&:480I6,/0<I ,8,20708=I 9/>60I##*II:9@0;I

supplies,andinstallationrails.IncludesChassisManagement;Upgradeableto

,-;4.'I8=0;:;4<0I,-;4.I ,8,20;I@4=3I:>;.3,<0I91I,-;4.' I4.08<0

MIS5300Q6DNC +

-/18884(1"'82,#*$

8D84&.,60

I)I'-<I$%I8D84,8/I3,<<4<I&@4=.3I<>::9;=482I>:I=9II:9;=<I

8.6>/0<II&:480I6,/0<I ,8,20708=I 9/>60I##*II:9@0;I<>::640<I

andinstallationrails.IncludesChassisManagement;UpgradeabletoFabricIT

8=0;:;4<0I,-;4.I ,8,20;I@4=3I:>;.3,<0I91I,-;4.' I4.08<0

MIS5600Q10DNC +

$#2,# ,1.1(-, *8 , &$+$,18-#2*$

PPC460ManagementmodulefortheMIS5xxxSeriesChassisSwitch,RoHSR5

MIS5600MDC +

$ %8* #$8%-/81'$85558$/($08' 00(084(1"'

8D84&.,60

I)II:9;=I$&#I-<I8D84,8/II

MIS5001QC +

,7, ,#8-018' ,,$*8# .1$/08 $** ,-58 /18 MFG

/#$/ !*$8

SinglePortQDRConnectX

I)#I,/,:=0;I.,;/I$&#II-<I,8/I42I

#0IAI'<I=,66I-;,.50=I%9&I%

$*'% +

DualportQDRConnectX

I)#I,/,:=0;I.,;/I$&#II-<I,8/I42I#0I

AI'<I=,66I-;,.50=I%9&I%

$*'% +

6,/008=0;II 0CC,8480I/,:=0;I,;/III,8/II-<I8D84,8/I>,6I#9;=I

#0I,<0II.97:64,8=II.97:,=4-60FI'<I9;I'<I6485I;,=0IA

Custom 46M6001

1'$/,$18# .1$/08 $** ,-58 /18 MFG

/#$/ !*$8

>,6#9;=I42I/,:=0;I.,;/I9880.=*

I!I80=@9;5I48=0;1,.0I.,;/I/>,6:9;=I

&#I#0IAI'<I=,66I-;,.50=I%9&I%

!#*'% +

-%14 /$8("$,0$ $** ,-58 /18 MFG

/#$/ !*$8

MellanoxFabricITMgrPerNodeRTU

,-;4.'

*SeeContacts

below

IS5100, IS5200, IS5300, and

IS5600 108, 216, 324 and 648-port

40Gb/s InfiniBand Modular Switch Systems

ConnectX-2 VPI Dual Port QSFP

InfiniBand and SFP+ 10GigE Adapter

ConnectX-2 EN Adapter Card with SFP+

Body Text

Univers LT 47 CondensedLt

8.5 pt/1 0 pt

Tracking: -5 pt

Table Text

Univers LT 47 CondensedLt

8.15 pt/9 pt

Tracking: -3 pt

Contact Information

Tables

Tables can be arranged

in one large column or

two columns with an

even width as shown.

Heading

Univers LT 55, Bold

11.5 pt

PMS 274 C

N$09;403N%0@9E0GN(C8B4N

(C<<GD0:4NN

)4:NNKN0FN

www.mellanox.com

REFERENCE GUIDE

MN=>G@867BNN"4::0<=FN)427<=:=684AN::N@867BAN@4A4@D43

"4::0<=FN@8364-N=<<42B-N$'8@42BN<I<8:0ABN<I<8@8364N<I<8=ABN<I<8'(N<I<8(20:4N<I<8%N%7G-N

+8@BC0:N%@=B=2=:N<B4@2=<<42BN0<3N+=:B08@4N0@4N@468AB4@43NB@034;0@9AN=5N"4::0<=FN)427<=:=684AN!B3N01@82)N8AN0NB@034;0@9N

=5N"4::0<=FN)427<=:=684AN!B3N::N=B74@NB@034;0@9AN0@4N>@=>4@BGN=5NB748@N@4A>42B8D4N=E<4@A

!' '""

7C29N)G1C@

DirectorofSales

)4:N

;08:N27C29;4::0<=F2=;

#!!!'$"'""

SteveWilliams

$"NCA8<4AAN4D4:=>;4<BN"<6@

4::N

;08:NAB4D4;4::0<=F2=;N

' "'""

DarrinChen

(@N8@42B=@N$"NN70<<4:N84:3N"9B6

4::N

;08:N30@@8<;4::0<=F2=;

# "'""

GlennChurch

84:3N>>:820B8=<AN<6

4::N

;08:N6:4<<;4::0<=F2=;

!'&'%"! SKU#

N%(*NA834NB=N2=<<42B=@NA834N08@J=EN(B0<30@3N34>B7N'08:N 8BNNN&(%NN%=E4@N(C>>:G

<2:C34AN70AA8AN"0<064;4<BN*>6@03401:4NB=N01@82)N<B4@>@8A4N01@82N"0<064@N

E8B7N>C@270A4N=5N01@82)"N!824<A4

(N<I<8(20:4LN+N&'N<I<80<3N(E8B27N*<;0<0643N



(N<I<8(20:4N+N&'N<I<80<3N(E8B27N"0<0643N%%-'



(N<I<8(20:4N+N&'N<I<80<3N(E8B27N"0<0643N%%-



(N<I<8(20:4N+N)1AN&'N70AA8AN(E8B27NAC>>BNC>NB=NN

ports**



(N<I<8(20:4N+N)1AN&'N70AA8AN(E8B27NAC>>BNC>NB=NN

ports**



(N<I<8(20:4N+N)1AN&'N70AA8AN(E8B27NAC>>BNC>NB=NN

ports**



<I<8(20:4N+NN>=@BN&(%N1AN<I<80<3N!405N:034N5=@NB74N

MIS5xxxSeriesChassisSwitch

A3000900

")(N<I<8(20:4N+N)1AN&'N<I<80<3N70AA8AN(E8B27NN:405N

A:=BANN(>8<4N:034AN"0<0643N%%-NN>=E4@NAC>>:84AN'=(N'N

<AB0::N@08:ANN@4?N:0<9AN8<2:C343N



<I<8(20:4N+NN>=@BN&(%N1AN<I<80<3N!405N:034N5=@NB74N

MTS3610SeriesChassisSwitch,RoHSR5,Spare



"&N1AN<I<80<3N70AA8AN(E8B27N5=@N"4N502B=@GN8<AB0::43



"&N1AN<I<80<3N70AA8AN(E8B27N5=@N"4NCABN 8B



!'&'" ! SKU#

N1AN0<3N86N%4NFN)ANB0::N1@0294BN'=(N'

"&-)'N=<<42B-LN+%N030>B4@N20@3NA8<6:4>=@BN&(%



"&-)'N=<<42B-N+%N030>B4@N20@3N3C0:>=@BN&(%



"/-)'N=<<42B-N+%N030>B4@N20@3N1ANN&(%N0<3N86N

(%N%4NFN)ANB0::N1@0294BN'=(N'



N.40@NFB4<343N,N,0@@0<BGNN30>B4@A

N

1AN<I<80<3N"4HHN20@3N5=@N"4N502B=@GN8<AB0::43N

4303431

1AN<I<80<3N"4HHN0@3N5=@N"4NCABN 8BN



!'&'%"! SKU#

N%=@BN<I<8(20:4N+N&(%N1ANN(E8B27



FBN,0@@0<BGN(E8B27NNN.40@



1AN<I<80<3N70AA8AN(E8B27N502B=@GN8<AB0::43



1AN<I<80<3N70AA8AN(E8B27NCABN 8B

NNNNN

!'&'" ! SKU#

N1AN0<3N86N%4NFN)ANB0::N1@0294BN'=(N'

=<<42B-LN+%N030>B4@N20@3NA8<6:4>=@BN&(%N



=<<42B-N+%N030>B4@N20@3N3C0:>=@BN&(%N



=<<42B-N+%N030>B4@N20@3N3C0:>=@BN-N



N.40@NFB4<343N,N,0@@0<BGNN30>B4@N0@3AN



1AN<I<80<3N"4HHN20@3N502B=@GN8<AB0::43



1AN<I<80<3"4HHN0@3NCABN 8B



Cables SKU#

1MPassiveCopper 

"N%0AA8D4N=>>4@ 

3MPassiveCopper 

4MPassiveCopper 

5MPassiveCopper 

6MPassiveCopper 

"N%0AA8D4N=>>4@ 

"N2B8D4N=>>4@ 

10MActiveCopper 

"N2B8D4N=>>4@ 

Cables SKU#

10MFiber 

15MFiber 

"N814@ 

30MFiber 

'"'" "'" ' ! SKU#

%4NFN)ANB0::N1@0294BN'=(N'

=<<42B-LN#N<4BE=@9N8<B4@5024N20@3N3C0:>=@BN(-



=<<42B-N#N<4BE=@9N8<B4@5024N20@3N3C0:>=@BN(%N



=<<42B-N#N(-N%4N)ANN>=@BA



=<<42B-N#N86N(%N%4NFN)ANN>=@BA



"4::0<=FN(%N=>B820:N;=3C:4N5=@N(!'



"4::0<=FN(%N=>B820:N;=3C:4N5=@N(('





 ''"  !' '" SKU#

Allincludesubnetmanagerandadministrator,clusterdiagnosticsandperformance

manager.Alsoincludes1yearsoftwaremaintenanceagreementforminorfeature

C>30B4AN0<3N1C6NIF4A

FabricITLicense(upto36ports)



FabricITAnnualMaintenance(upto36ports)



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82N)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



FabricITLicense(upto144ports)



FabricITAnnualMaintenance(upto144ports)



01@82)N!824<A4NC>NB=NN<=34A



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



01@82)N!824<A4NC>NB=NN>=@BA



01@82)N<<C0:N"08<B4<0<24NC>NB=NN>=@BA



3.375”

CORPORATE STYLE GUIDE

19 | Mellanox Technologies | July 2011

Main Title

Univers LT 59 UltraCondensed

26 pt / 29

Documents and Marketing Collateral

TECHNOLOGY BRIEFS

TECHNOLOGY BRIEF

The two commonly known RDMA (remote DMA) technologies are InniBand and iWARP (Internet

Wide Area RDMA Protocol). InniBand has enjoyed signicant success to date in HPC applications.

iWARP solutions over Ethernet have seen limited success because of implementation and deployment

challenges. Recent enhancements to the Ethernet data link layer under the umbrella of IEEE data center

Bridging (DCB) open signicant opportunities to proliferate the use of RDMA technology into mainstream

data center applications by taking a fresh and yet evolutionary look at how such services can be more

easily and efciently delivered over Ethernet. The proposed DCB standards include: IEEE 802.1bb –

Priority-based ow control, 802.1Qau – Congestion Notication, and 802.1az – Enhanced Transmission

Selection (ETS) and DCB Capability Exchange. The lossless delivery features in DCB, enabled by Priority-

based Flow Control (PFC), are analogous to those in the InniBand data link layer. As such, the natural

choice for building RDMA services over PFC-based DCB Ethernet is to apply use InniBand-based native

RDMA transport services. The IBTA (InniBand Trade Association) has recently released a specication

called RDMA over Converged Ethernet (RoCE, pronounced as “Rocky”) that applies the InniBand-based

native RDMA transport services over Ethernet. ConnectX-2 with RoCE (RDMA over Ethernet) implements

the RoCE standard to deliver InniBand-like ultra low latency and high scalability over Ethernet fabrics.

ConnextX-2 with RoCE is born out of combining InniBand native RDMA transport with Ethernet per the

IBTA RoCE specication. The data link InniBand-based layer 2 is replaced by Ethernet layer 2, as shown

in the gure below. The InniBand transport is applied over a PFC-based loss less Ethernet data link.

LRH

(L2 Hdr)

GRH

IB Transport

Headers

IB Payload

ICRC

VCRC

InfiniBand

RoCE

Eth L2

Header

GRH

IB Transport

Headers

IB Payload

ICRC

FCS

Application

IB Software transport

interface

IB-based transport

Network

Ethernet w/PFC (Data Link)

OFA verbs

Figure: Low Latency Ethernet packet

format and protocol stack

Software Interface: ConnextX-2 with RoCE is compliant with the Open Fabrics Alliance OFED

verbs denition and is interoperable with the OFA software stacks (similar to InniBand and iWARP).

ConnextX-2 with RoCE uses the proven and feature rich InniBand verbs interface available in the OFA

stacks. OFED v1.5.1 supports RoCE and ConnextX-2 with RoCE.

Transport Layer: ConnextX-2 with RoCE uses the InniBand transport layer, as dened in the IBTA RoCE

specication. The adaptation from InniBand data link to Ethernet data link is straight forward because

the InniBand transport layer was designed ground up to be data link layer agnostic. The InniBand

transport layer expects certain services from the data link layer related to lossless delivery of packets, and

these are delivered by a PFC enabled Ethernet data link layer. ConnextX-2 with RoCE inherits a rich set

of transport services beyond those required to support OFA verbs including connected and unconnected

modes, reliable and unreliable services. Built on top of these services is a full set of verbs-dened

operations including kernel bypass, send/receive, RDMA read/write, and atomic operations.

1.0 Opportunities with

Evolution of Ethernet

1.1 How ConnectX-2

RoCE Works

ConnectX

-2 with RoCE

(ConnectX-2 VPI and ConnectX-2 EN)

April 2010

. &$;

TECHNOLOGYBRIEF:

ConnectX

2withRoCE

350OakmeadParkway,Suite100,Sunnyvale,CA94085





:;-.70(&'2;;$** ,-6;$"',-*-&($1;**;0(&'21;0$1$04$#

$** ,-6;0(#&$;-,,$"2;(0$"2;,9,(* 12;,9,(0(#&$;,9,(-12;,9,(;,9,(" *$;,9,(;'7;(023 *;0-2-"-*;,2$0"-,,$"2; ,#;-*2 (0$; 0$;0$&(12$0$#;

20 #$+ 0)1;-%;$** ,-6;$"',-*-&($1;2#; !0(";(1; ;20 #$+ 0);-%;$** ,-6;$"',-*-&($1;2#;**;-2'$0;20 #$+ 0)1; 0$;.0-.$027;-%;2'$(0;0$1.$"2(4$;-5,$01

Management Feature Required by

IB Transport Layer & Apps using

IB Transport Layer

How InﬁniBand delivers them in

the InﬁniBand subnet

How Ethernet (and CEE) delivers

them using standard Ethernet

management practices

L2 address assignment Subnet Manager L2 address

assignment

Fixed assigned L2 address or other

Ethernet mechanisms

L2 topology discovery and switch FDB

conﬁguration

Subnet Manager topology discovery

using direct routed subnet manage-

ment packets (SMP). Subnet Manager

path computation and path distribution

Spanning Tree and Learning mecha-

nisms. Also, IEEE Transparnet Intercon-

nection of Lots of Links (TRILL) when

available and other eth practices

Address Resolution SA based path resolution Address Resolution Protocol (ARP) or

direct mapping

QoS QoS Manager extension to Subnet

Manager

Standard Ethernet QoS management

practices. Local API to access fabric

policy settings

Congestion management Congestion Manager for IB 802.1Qau congestion management

features

Performance management IB Performance Manager SNMP/RMON MIBS

Device/baseboard management IB Baseboard Manager SNMP/RMON MIBS

-,,$62;5(2';-; # .2$01;! 1$#;-,;2'$;;-;1.$"(9" 2(-,; 0$; 4 (* !*$;2-# 7;%0-+;$** ,-6;

$"',-*-&($1; ,#;' 4$;!$$,;#$+-,120 2$#;2-;#$*(4$0;$,#;2-;$,#; ..*(" 2(-,;*$4$*;* 2$,"($1;-%; 1;*-5; 1;

;+("0-1$"-,#1;$** ,-6; ,#;-2'$0;(,#31207;*$ #$01; 0$;"-** !-0 2(,&;-,;&0-5(,&;2'$;$"-1712$+;-%;

-! 1$#; # .2$01; ,#;(,#$.$,#$,2;1-%25 0$;4$,#-0; ..*(" 2(-,1;2' 2;" .(2 *(8$;-,;2'$;!$,$921;-%;

-,,$62;5(2';-;-+$;$6 +.*$1;-%;2 0&$2; ..*(" 2(-,1; 0$;9, ,"( *;1$04("$1;!31(,$11;(,2$**(&$,"$;

# 2 ;5 0$'-31(,&;"*-3#;"-+.32(,&; ,#;$!;

 1$#;-,;2'$;#(1"311(-,; !-4$;(2;(1;-!4(-31;2' 2;-,,$62;5(2';-;"-+$1;5(2';+ ,7; #4 ,2 &$1;

,#;'-*#1;2'$;.0-+(1$;2-;$, !*$;5(#$1.0$ #;#$.*-7+$,2;-%;;2$"',-*-&($1;(,;+ (,120$ +;# 2 ;

"$,2$0; ..*(" 2(-,1;;

1. -,,$62;5(2';-;32(*(8$1; #4 ,"$1;(,;2'$0,$2;;2-;$, !*$;$%9"($,2; ,#;*-5;"-12;(+.*$

+$,2 2(-,1;-%;;-4$0;2'$0,$2;;

2. -,,$62;;20 %9";" ,;!$;"* 11(9$#; 2;2'$;# 2 ;*(,);* 7$0;5'("';(1;% 12$0; ,#;0$/3(0$1;*$11;

;-4$0'$ #

3. -,,$62;5(2';-;#$*(4$01;31$"; ..*(" 2(-,;2-; ..*(" 2(-,;* 2$,"7;5'("';(1;2';-%;-2'$0;

(,#31207;12 ,# 0#;(+.*$+$,2 2(-,1;-4$0;2'$0,$2;$,"'+ 0)(,&;5(2';.-.3* 0;9, ,"( *;1$04("$1; .

.*(" 2(-,1;1'-5;+-0$;2' ,;;*-5$0;* 2$,"7; ..*(" !*$;2-;" .(2 *;+ 0)$2;# 2 ;.0-"$11(,&; ,#;20 #$;

$6$"32(-,1

4. -,,$62;5(2';-;13..-021;2'$;$,2(0$;!0$ 2';-%;; ,#;*-5;* 2$,"7;%$ 230$1;'(1;(,"*3#$1;

0$*( !*$;"-,,$"2$#;1$04("$;# 2 &0 +;1$04("$;; ,#;1$,#0$"$(4$;1$+ ,2("1; 2-+(";-.$0 2(-,1;

31$0;*$4$*;+3*2(" 12;31$0;*$4$*;; ""$11;)$0,$*;!7. 11; ,#;8$0-;"-.7;;

5. '$;;4$0!1;31$#;!7;-,,$62;5(2';-; 0$;! 1$#;-,;,9,( ,#; ,#;' 4$;!$$,;.0-4$,;(,;

* 0&$;1" *$;#$.*-7+$,21; ,#;5(2';+3*2(.*$;; ..*(" 2(-,1;!-2';(,;2'$;; ,#;;1$"2-01;3"';

..*(" 2(-,1;" ,;,-5;!$;1$ +*$11*7;-%%$0$#;-4$0;-,,$62;5(2';-;5(2'-32; ,7;.-02(,&;$%%-02;

0$/3(0$#

6. -,,$62;5(2';-;! 1$#;,$25-0);+ , &$+$,2;(1;2'$;1 +$; 1;2' 2;%-0; ,7;2'$0,$2; ,#;

! 1$#;,$25-0);+ , &$+$,2;$*(+(, 2(,&;2'$;,$$#;%-0;;+ , &$01;2-;*$ 0,;,$5;2$"',-*-&($1

1.2ConnectX2with

RoCEAdvantages

Body Heading

Univers LT 47 CondensedLt, Bold

12 pt/13 pt

PMS 274 75%

Set as Inline Heading anchored to

beginning of paragraph

Main Title 2

Univers LT 47 CondensedLt

15 pt /15

.75”

.5”

2.625”

CORPORATE STYLE GUIDE

20 | Mellanox Technologies | July 2011

Documents and Marketing Collateral

WHITE PAPERS

WHITE PAPER

2010

The security trading market is experiencing rapid growth in volume and complexity with a greater reliance

on trading software, which is supported by sophisticated algorithms. As this market grows, so do the

trading volumes, bringing existing IT infrustracture systems to their limits.

Middleware and networking solutions must not only deliver lower latency despite higher volumes,

but also sustain performance through market data trafc spikes while minimizing latency jitter and

unpredictable messaging outliers.

To address this challenge, two market leaders have come together to provide a eld-proven

infrastructure solution that combines best of breed enterprise messaging middleware and trading

transport infrustructure. The solution slashes end-to-end trading latency by analyzing each layer of

the infrastructure—server, storage, server stack and middleware, as well as switching and bridging

devices—and optimizing each part for minimum latency and maximum throughput, while keeping an eye

on overall cost.

29West has emerged as the standard in next generation messaging software and delivers an unbending

commitment to offering the lowest possible latency across the widest possible range of messaging

use cases. Similarly, Mellanox has evolved into the market leader in IT scale-out fabric infrastructure,

excelling in low latency software and switching solutions.

This report demonstrates the benets of a 29West middleware solution integrated with Mellanox’s low-

latency solution for high-frequency trading customers. Specically, this paper describes a benchmark of

the integrated solution that delivers application messaging at:

Latency under medium load

Message rates under 125K msg/sec, message size under 256 bytes

• Average latency under 9 µsec

• 99.9th percentile latency 45 µsec or lower

Latency under heavy load (high throughput)

Message rate up to 1,300,000 msg/sec, message size under 64 bytes

• Average latency under 18 µsec

• 99.9th percentile latency 70 µsec or lower

Executive Summary

Executive Summary ......................................................................................................................1

Reference Transport Latency .......................................................................................................2

Latency Versus Message Size .....................................................................................................3

Conclusion ....................................................................................................................................5

Methodology of Measurement ....................................................................................................5

Benchmark Set Up Details ...........................................................................................................6

About 29West Inc,. ......................................................................................................................8

A Best of Breed Low Latency Solution for Trading

Environments from 29West & Mellanox

29West Latency Busters

Messaging (LBM) and Messaging

Accelerator

™

(VMA) Over Mellanox 40Gb/s (QDR) InniBand Fabric

01 TOC Listing

Univers LT 47 CondensedLt

10 pt/13

Generated automatically using

01 Body Heading1 tags.

Page is loading ...

Mellanox Technologies MC2210130-003 Datasheet

Brand: Mellanox Technologies

Model: MC2210130-003

Type: Datasheet

This manual is also suitable for

Download PDF

report

Mellanox Technologies MC2210130-003 Datasheet

This manual is also suitable for

Mellanox Technologies MC2210130-003 Datasheet

Related papers

Other documents