 |
|
DAILY NEWS AND INFORMATION FOR THE GLOBAL GRID COMMUNITY /
|
Breaking News - Networking:
InfiniCon Adds To InfinIO 9000 Series For Enterprise Grids
InfiniCon Systems announced new members of the InfinIO 9000 Series -- the most
scalable, integrated switching platform for enabling large computing clusters
and enterprise data center Grids.
The 9000 Series consists of models offering up to 144 and 288 10Gb ports in
compact 7U and 14U chassis. It combines the industry's fastest standard
networking architecture -- InfiniBand -- with both Fibre Channel and Ethernet,
to make scale-out processing for large clustering applications more powerful
and affordable, and to bring the resource sharing and complexity reduction
benefits made possible through I/O virtualization to data center planning and
management.
The 9000 Series offers the following advances and key differentiators over
competitive solutions:
- Design Simplicity and Investment Protection Through Common Components --
Users can deploy scalable systems of either 144-port (9100 Model) or 288-port
capacities (9200 Model). The bladed systems architecture utilizes common
components across chassis sizes, enabling sensible migration strategies as
clusters scale. No other InfiniBand system provides this level of investment
protection.
- Cost-Effective Integration into Enterprise Data Centers -- To equip servers
with shared access to resources such as Fibre Channel SANs, NAS appliances and
client networks, 9000 Systems can be configured with virtual I/O blades, which
also contain InfiniBand ports. Users can determine what level of virtual I/O
capability is needed, without sacrificing port counts that impair InfiniBand
fabric sizes. No other InfiniBand system offers this level of granularity.
- Unmatched Power for High Performance Computing Applications -- A fully
configured 288-port InfinIO 9200 delivers an incredible 5.76 Tbits per second
of bidirectional throughput, a performance metric capable of matching the most
demanding computing applications. InfiniBand blades supporting 30Gb networking
can be mixed and matched within the 9000 Series chassis, enabling ultra high
performance uplinks required for configuring larger fabrics.
"The industry is moving to clustered, fabric, and Grid computing architectures
which require at their very foundation large-scale switching systems," stated
Charles Foley, executive vice president of InfiniCon. "Our 9000 Series is the
most powerful data center switching system available today to enable this
strategic shift, and the first to combine ultra scalable, high-performance
switching with uncompromised access to storage and LAN/WAN."
InfiniCon offers the industry's most complete family of InfiniBand solutions
that target mid-to-high end HPC applications through solutions like the
InfinIO 9000, as well as smaller HPC configurations and clustered database
applications through solutions like its new 12-port InfinIO 5000 Clustering
and I/O Virtualization System.
"We clearly see the move towards 'scale out' computing gaining momentum, and
have capitalized on InfiniBand as a strategic technology in this evolution,"
stated Vijay Agarwala, director of high performance computing at Penn State
University. "The new systems from InfiniCon are perfect examples of what are
not only logical, but necessary, developments to enable this continued
migration."
Full High Availability and Robust Management Environment
Every aspect of the 9000 Series is designed to meet the rigors of
mission-critical environments. High availability features include redundant
power/cooling, hot pluggable components, redundant management, and full
failover. It offers complete IBTA 1.1 management support, world-class subnet
management tools such as the unique FastFabric Toolset, and intelligent
balanced routing -- an industry exclusive feature for HPC environments that
directs traffic across optimal paths rather than saturating interswitch links
like competitive design approaches do, risking application bottlenecks.
"Systems such as the InfinIO 9000 will be cornerstones in the broad industry
move to fabric/Grid/cluster computing," stated Jamie Gruener, senior analyst
with the Yankee Group. "The performance and scalability make commodity 'scale
out' processing viable, and the integration of fibre channel and Ethernet
allow seamless integration into the existing environment; both are key
criteria for acceptance of new infrastructure building blocks such as this."
InfiniCon is a member of Sun Microsystems' High Performance Technical
Computing Alliance, and also partners with AMD, Fujitsu, Intel, Oracle and
other leading IT companies.
|