Iron provides a turn-key, high performance computing solution which combines the flexibility of a custom solution with the simplicity, reliability, availability, compatibility, power efficiency and value to meet your application and budget needs. Iron has introduced platforms based on scalable cluster concept to create next-generation clusters with supercomputer performance. The concept is based on a flexible architecture that groups together clusters to make a unified, fully integrated scalable platform that can be provisioned and managed as a stand-alone HPC system to deliver sustained capacity computing performance.

Our solution includes:

  • Custom Packaging: Choice of small to large scale turn-key cluster packaging solutions which includes:
  • Wokrgroup Cluster
  • High Density Integrated Rack Cluster
  • High Density Modular Rack Cluster, expandble modular design expanding from 10's to 100's of nodes

 

  • Custom HPC Configurations:
  • Custom Hardware Configuration: Choice of processors, system configuration, cluster interconnects and middleware
  • Custom Software Configuration: Choice of operating system, software and tools

 

  • Turnkey Advantage: Ready to Deploy, All the hardware and software comonents are factory integrated, tested and verified

Custom Packaging: Turnkey Cluster Packaging Options

Workgroup Cluster:
Workgroup Cluster solution is available as a standalone rackmount on wheel. This is a self contained "all-in-one" turnkey solution with upto 14+ powerful servers, interconnects, a master node, optional storage, power and coolling management, OS, middleware and cluster management software. Workgroup Cluster provides larger and more complex computer-intensive workloads outside of the data center with limited IT support, its ideal for midsize business with limited budget.

High Density Integrated Rack Cluster:
Rack Cluster is available in 1, 2, 3 or 4 rack cabinet onfigurations. Each of the racks are optimized to support over 32 1U servers or a large number of blade servers. A high density 4 rack cabinet system is available with 128 1U servers.

High Density Modular Rack Cluster:
Modular Rack Cluster option helps build a highly scalable HPC solution, it seperates the computing resources from storage, interconnect and console tasks. All of the compute nodes are integrated into dedicated computing rack cabinets, interconnect rack cabinet contains the switching gear, all of the reaminder hardware blocks including management gear, maste node, storage, keyboard, monitor are integrated into a seperate rack.

Custom HPC Configurations: Multiple Node Types, Highly Scalable Performance

Node Type High Density Blade Servers Twin Server Quad Processor Server
 
Rack Capacity (44U)      
Total Systems 60 80 40
Total Processors (Cores) 240 (960) 160 (640) 160 (640)
Node Configuration Blade Servers, 7U Twin 1U Server Quad 1U Server
Processor Family Intel Xeon or AMD Opteron Intel Xeon or AMD Opteron Intel Xeon or AMD Opteron
Socket Count (Cores) 4 (Quad), 64-bit 2x2 (Quad), 64-bit 4 (Quad), 64-bit
Cores Per Node 16 16 16
Memory per Node (Max) 128GB 2x64GB 64GB
Hotswap Disk Drives Yes Yes Yes
Hotswap Power Supplies Yes Yes Yes
Hotswap Cooling Yes Optional Optional
PCI Express Slots No Yes Yes
Cluster Interconect Infiniband, GbE with Myrinet and Quadrics available on non-Blade configurations
System Management Console & Admin Network thru GbE Ethernet based IPMI connection
Software Linux (Red Hat EL or SuSE), Microsoft Windows Server 2008 and more
Rack Cabinet Managemnet      
Serial Console Management Optional Optional Optional
Power Distribution Unit Yes Yes Yes
UPS Optional Optional Optional
Cluster Managers Linux Beowulf, Microsoft HPC Server, Moab Cluster Suite

Turnkey Advantage: Rack Optimized

Proven Technology:

  • Support for Quad-Core Intel® Xeon® processors and AMD® Opteron® processors
  • Intel® Cluster Ready Certified Platforms
  • Advanced Intel compilers and development tools
  • Fully integrated, pre-packaged solution with a variety of system configurations

System Scalability:

  • 160 sockets with 640 cores and 4TB of memory per rack
  • Scale to 6TF in a single 44U rack

System Reliability:

  • Improve reliability, availability and serviceability
  • Redundant (Dual-Rail) Infiniband Interconnect Fabric Option
  • Redundant (Dual-Rail) Ethernet Management Fabric Option
  • Independently controlled cooling fans per node
  • Redundant N+1 power supplies, highest efficiency
  • Advance Rack Cooling:
  • Improve power and cooling efficiency
  • Optimized system efficiency reduces cooling costs with improved MTBF by up to 10X


System Management:

  • Comple lights-out management system