SUPerG 2005 Paper
By dcb on Feb 22, 2005
SUPerG (Sun Users Performance Group) is Sun's premiere technical event for customers interested in large scale solutions architected for data centers and high-performance and technical computing. The program is designed to provide highly interactive and intimate exchanges between Sun's leading technical experts and our customers.
You can read more about the event, and register to attend at: http://www.sun.com/datacenter/superg/
This year, I've been invited to speak at this event. In the spirit of blogging, I've posted my abstract (below). I need to get busy writing the paper and creating a clear, concise, and compelling technology demonstration. If you'd like me to address a particular topic or concern or challenge in the paper (related to the abstract), or have an idea that I might include in the demo, please drop me a line or submit a comment to this entry. If I use your idea, I'll attribute it to you in the published paper, and here in blog land, so please include your name and contact info.
Hope you see you at SUPerG in April in Virgina. Stop by and say "hi".
\*SUPerG 2005 Abstract\*
Effective Deployment & Operational Practices for Dynamically Scaled SOA and Grid Environments
Scalability is taking on a new form and introducing new challenges. The traditional "vertical" platform with dozens of powerful CPUs bound by local memory offering (nearly) uniform memory access, is being threatened by a new model - networked grids of powerful but low cost compute nodes.
Grids are not new. But powerful new techniques are emerging that allow commercial workloads to take advantage of this style of computing. This includes SOA-based application design, as well as auto-deployment and provisioning to drive efficiency and utilization in infrastructure operation.
Modern designs provide for on-the-fly horizontal scaling with the push of a button.... in which new containers join the grid into which a distributed app may expand to offer new levels of performance and service level. A side effect of this approach is a highly-resilient platform, such that bound dependencies can fail without a catastrophic impact on the running service.
This talk will provide an update on the State of the Technology with respect to SOA and Infrastructure Provisioning, and how these can be leveraged to offer Adaptable, Scalable, and Resilient services.
I may also include a demonstration that will show how a collection of bare metal servers can be established into a Grid using N1 SPS (integrated with JET). Following this provisioning phase, the demo will then show a sample app deployed and executed across multiple nodes. Finally, it'll show a node being added to the live Grid using SPS, and how that app can then expand, at run-time, to leverage this new node, increasing its work rate.