Close the Cluster Smorgasbord

September 15th, 2010 11:19 am
Posted by Brock Taylor
Tags: , , , , , , ,

I expect there will always be custom-built cluster systems that target proprietary software applications or specialized environments.  Those systems are built by experts for experts and just about every design decision is painstakingly made.  HPC clusters for the masses will not be able to go down that path.

Volume HPC cluster solutions need to move away from cafeteria selection of components to pre-defined, flexible designs that meet a broad spectrum of user needs. This will take a lot of complexity out of purchasing HPC clusters and remove the need for the customer to be a cluster systems expert. I'm not suggesting that cluster purchasers have zero understanding of the system, but it needs to be simplified to understanding the application workload needs that will run on the system.

Purchase time decisions for volume cluster solutions need to center on aspects that affect execution in terms of performance but do not risk introduction of component integration issues.  That means that components like operating system, cluster provisioning, interconnect hardware, and resource management software need to be set and unchangeable at purchase time.

Allowing total piecemeal selection of components at purchase time requires the buyer to know how everything fits together, have the tools and knowledge to build the system, and/or pay for the services to put it together.  This approach works for veteran buyers who easily justify the return on investment but is much harder for the (new) volume cluster buyer.

Volume cluster solutions are more like partnerships of many components with the solutions provider as the focal point.  Solidify the solution component partnership, and the solutions themselves become easier to produce and ultimately become more stable product. That helps lower barriers to using HPC clusters.

There's a key set of partners I haven't covered yet, though, the application vendors themselves, but that's another topic all together.

>> See Related Stories for more about how to simplify selecting, buying, and managing HPC clusters... and come back for more from this weekly blog.


You must be a Registered Member in order to comment on Cluster Connection posts.

Members enjoy the ability to take an active role in the conversations that are shaping the HPC community. Members can participate in forum discussions and post comments to a wide range of HPC-related topics. Share your challenges, insights and ideas right now.

Login     Register Now

Author Info
Brock Taylor

Brock Taylor is an Engineering Manager and Cluster Solutions Architect for volume High Performance Compute clusters in the Software and Services Group at Intel. He has been a part of the Intel® Cluster Ready program from the start, is a co-author of the specification, and launched the first reference implementations of Intel Cluster Ready certified solutions.

Brock and others at Intel are working within the HPC community to enable advances and innovations in scientific computing by lowering the barriers to clustered solutions.

Brock joined Intel in December of 2000, and in addition to HPC clustering, he previously helped launch new processors and chipsets as part of an enterprise validation BIOS team. Brock has a B.S. in Computer Engineering from Rose-Hulman Institute of Technology and an M.Sc. in High Performance Computing from Trinity College Dublin.