High Performance Computing
“When we purchased our cluster from Alliance Technologies it increased the compute capacity for our Engineering Group by an order of magnitude – simply put, we can perform simulations that were just not possible before. Alliance Technologies has proven to be a valuable partner in supporting this critical resource; not only in maintaining the hardware, but also ensuring that it remains compatible with code updates and operating at peak performance.”
Engineering Manager, Automotive Components Manufacturer, Strathroy, Ontario
Alliance Technologies are experts in High Performance Computing (HPC). Whether you need hardware to host a virtual infrastructure, an application specific compute server or cluster or, a server to run a custom Linux OS, we have the expertise to deliver the compute performance you need within your budget.
Our turn-key solutions include:
- Virtual host servers
- High Performance Compute Servers
- Compute Clusters and Grid Computing
Contact us to arrange a no obligation assessment of your compute needs with one of our HPC professionals.
Virtual Host Servers
To maximize the benefits and performance of your virtual infrastructure you need the right host hardware. Alliance Technologies understands virtualization and can help you select the right host servers and other complimentary components to meet your needs.
Fully compatible with VMware and with high availability features and configuration options such as "boot from SAN", choose a custom virtual host server from F4 Labs or class-leading hardware from Dell or HP.
Ask us about having VMware VI3 pre-loaded as a no charge option!
F4Labs offer a range of servers taylor made for your specific host infrastructure requirements; whether they be for an all-in-one host for a back-up virtual infrastructure, a cost effective replication solution, or a state of the art "boot from SAN" server to host your core production infrastructure.
The F4Labs SaRA (Storage and Replication Appliance) is specifically designed for the small to mid size enterprises that need a cost effective replication solution. Fully compatible with VMware, and with a full range of drive options for either capacity or I/O performance, F4Labs SaRA can be deployed as a centralized storage solution to a VMware infrastructure or, as a replication gateway.
For smaller organizations or workgroups, F4Labs SaRA is available with sufficient compute power to be a complete, cost-effective hardware solution to host a virtualized infrastructure.
F4Labs SaRA can also be used in conjunction with the Business Continuum service, bringing the benefits of a full disaster recovery site and infrastructure to those organizations that do not have a secondary site. Contact us today to learn more about F4Labs SaRA and find out how you can implement a cost effective replication strategy.
Where the virtual infrastructure requires greater compute power, the F4Labs offer specialized Virtual Host versions of their Thunder and Fury compute servers -- the F4 Labs Thunder/VI and F4 Labs Fury/VI. These servers have the power and capacity to provide an extremely high virtual machine density -- a full populated F4Labs Thunder/VI can potentially host over 150 virtual servers -- making them ideal for use with a high performance centralized SAN to provide the full High Availability benefits of VMware.
To meet the wide range of requirements presented by our clients, Alliance Technologies recommends:
Ask us about having VMware VI3 pre-loaded as a no charge option on your new F4 Labs Servers!
Alliance Technologies are a fully authorized Dell partner, and can supply Dell’s full range of VMware compatible servers. Dell servers can be delivered with VMware pre-installed and are an ideal compliment to a Dell EqualLogic iSCSI SAN. Dell’s range of servers include two and four cpu servers with Intel or AMD quad-core processors, as well as their high density dual Socket, dual-core AMD or Intel Blade Server.
To meet our client’s most demanding compute applications, Alliance Technologies offer high performance, high value servers from F4 Labs. Each F4 Labs server is designed and configured specifically to meet your needs and budget -- allowing you to reduce hardware expenditure and maximize the return on your IT investment with maximum performance and reliability.
Compute servers from Alliance Technologies can be supplied with a choice of Operating System, including a full range of LINUX distributions, pre-installed.
F4 Labs Thunder
F4 Labs Thunder servers are built to provide ground-breaking performance for the most computationally demanding applications, such as numerical analysis, CAE or CFD solvers.
Configuration options include:
- CPU - Up to four (4) AMD Opteron Quad-Core processors (for 64 cpu cores)
- Memory - Up to 512 GB RAM
- Hard Drives - Up to eight hot-swap SAS or SATA drives optimized for swap or storage
- Power - Redundant PSU
- Chassis - Rack-mount or "Desk-Side".
F4 Labs Fury
F4 Labs Fury servers are available in a full range of configuration options to meet the specific needs of your applications and budget. Need more storage? F4Labs ServStor series provides storage (24TB or more) and the compute performance of the F4 Labs Fury all in one unit.
Compute Clusters and Grid Computing
Compute Clusters and Grid Computing is a core competency at Alliance Technologies. Compute Clusters designed and built by Alliance Technologies are in operation at both academic research labs and commercial organizations in the areas of Nuclear Physics, Engineering, Health Sciences, Aerospace and Astrophysics.
A cluster from Alliance Technologies will draw upon the high levels of research we undertake so that the latest in compute technologies can be incorporated. An example of this is the self contained, centrally cooled cluster delivered to the Department of Nuclear Physics at a major Ontario University:
Cluster Example - Centrally Cooled, Self Contained Cluster
Alliance was approached by the Department of Nuclear Physics at a major Ontario University, who wanted to add four or five compute nodes to their existing cluster. They also wanted to reduce noise levels from the cluster as the room in which it was located was to be used for additional office space -- an ideal application for Alliance Technologies centrally cooled, self contained cluster design. Therefore, rather than simply add compute nodes to the University’s existing cluster, Alliance Technologies, supplied a new system, with the latest CPU technology. We were also able to reuse the existing nodes by adapting them to fit into the new system. The result? The cluster was delivered on budget and delivered significantly more compute power that originally anticipated by the university -- with the added benefit of a dramatic reduction in noise levels generated by the cluster, quiet enough that the room can now be used as an additional workspace.
Many organizations there is no shortage of high performance hardware capacity. The problem is that much of this valuable resource is effectively inaccessible as the hardware is typically dedicated to specific applications even though full capacity on that hardware is rarely used. Consider also that when hardware is idle, software licenses -- that are often a much greater cost than hardware - are also not being used.
By intelligently implementing a Grid Computing infrastructure using PBS Professional, Alliance Technologies can dramatically improve the productivity and time-to-solution of HPC environments through increased 24/7 availability of computing resources.
PBS Professional reduces costs by maximizing the utilization of hardware and using software licenses more effectively. It reduces system administration costs by simplifying user access and providing more precise usage reports. Key features include:
- Robust, flexible workload management
- Industry-leading scalability with advanced, flexible queues
- Management and execution of many jobs as a single array
- A consistent interface across all platforms
- Scheduling choices for better workflow and utilization
- Resource-based scheduling
- Maximized utilization of solver licenses and hardware
- Pre-emptive scheduling for instant launching of critical jobs
- Fair-share scheduling and support for advanced reservations
- Industrial-grade reliability, availability and serviceability
- Server software failover for 24x7 availability of services
- Automatic detection of systems failures
- Automatic rescheduling of disrupted jobs
PBS Professional is available as a pre-installed option on all compute servers supplied by Alliance Technologies.
Contact us today to discuss how Alliance Technologies can help you increase the return on your HPC hardware and software investment.