I'm looking forward to attending TechEd in Boston. There are some great breakout sessions planned on Windows infratsructure technologies. So over the course of the next week I'll highlight some of them in case you haven't decided which sessions to attend. Today's post will focus on High Performance Computing.

But first, when I attended a session last week for TechEd speakers I was amazed to learn that 12,000 people are expected to attend the Boston show.  And that there are TechEd shows in 22 cities around the world with an expected reach of 25,000 customers. So I hope you can attend a show in your area.

Back to HPC. Here we're talking making massively-parallel computing accessible to Windows admins. It's no longer the realm of just national computing centers. I've been fortunate enough in the past few weeks to speak to several early adopters of Windows Compute Cluster Server. Several customers have literally given the reviewers guide and CDs to their Windows system admins and within a few hours they're installed and ready to schedule jobs. One thing they say is overlooked is that Windows CCS is not just an x64 operating system. The 2nd CD of Windows CCS (yes, there are 2 CDs if you get the physical media) contains a job scheduler, message passing interface, cluster monitoring tools and deployment tools. Add to that the OS integration to AD, MOM, use of RIS for unattended compute node installation and Visual Studio 2005 includes support for OpenMP standard and a parallel debugger for developing HPC apps.

I recently caught the following interview with Jay Boisseau, director of the Texas Advanced Computing Center (TACC), in an article in GridToday:

Anybody who thinks Microsoft's entry into HPC is not potentially a great thing is crazy, but, of course, [Microsoft] should obtain guidance and feedback from our community - and they are, via these partnerships.

So I recommend making yourself familair with Windows CCS. There are two sessions at TechEd (see below) and lots of demos in Microsoft's booth, to include 3rd-party apps for manufacturing, oil & gas, financial services and life sciences. Even if you don't get involved with HPC apps, you probably have groups within your company that do and they may not know that there's now an alternative to Linux-based HPC clusters. Or you may have people running HPC jobs on high-end workstations that could use some additional horse power from a server-based implementation.

 

Patrick

 

SVR203  Compute Cluster Server 2003: An Overview

Monday, June 12 10:45 AM - 12:00 PM, 156 ABC

Speaker(s): Kyril Faenov, Ryan Waite

Applications and scenarios for high performance computing (HPC) are increasing in automation, speed, and ease of use. Microsoft will deliver the first step towards Windows-based computational clustering with Windows Compute Cluster Server 2003 - a comprehensive, HPC solution surrounded by a broad ecosystem of partners, products and services to maximize business value. This session provides an overview of the CCS offering and highlights opportunities to leverage CCS within enterprise computing environments.

Session Level(s): 200

 

SVR325  Deploying and Managing a Windows-Based High-Performance Compute Cluster

Thursday, June 15 8:00 AM - 9:15 AM, 153 ABC

Speaker(s): Onur Koc

For customers solving complex computational problems, Windows Compute Cluster Server 2003 accelerates time-to-insight by providing a High-Performance Computing (HPC) platform that is simple to deploy, operate, and integrate with existing infrastructure and tools. This session provides a technical overview of Windows Compute Cluster Server 2003, and guidance on deploying and managing a Windows-based compute cluster.

Session Level(s): 300