Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

My Job is to Hallucinate: Unisys CTO

Eric Ernest | March 20, 2014
Jim Thompson, CTO, Unisys, talks to Computerworld India about the company's Forward! solution, the state of the Unix mission critical market and how India figures in Unisys's future plans.

So we believe that you need a uniform architecture around Xeon. We want to facilitate the movement of PA-RISC workloads that are seen as critical into the Forward! space. We believe that we have a unique capability in that most of the work we do today is centred around mission critical environments on our ClearPath mainframes. The activities we do in telecommunications, transportation, finance and public sector segments tend to be mission critical.

We think there is a huge opportunity to leverage this using Xeon technology and Linux technology, from our partners at Rad Hat and Suse and to deliver business critical environments that run on Linux for the PA-RISC world. We are working with SAP to deliver SAP ERP on the Forward! platform. We are also working on consolidation. So these are 3 use cases we are working on — consolidation, workload PA-RISC / Unix migration, and SAP.

Imagine a system that is comprised of standardized Intel building blocks where each node on a blade brings lot of I/O capability, a lot of memory capability. Now Blades tend to be limited in that sense.

So you band them together with a high speed low latency medium — we are using Infiniband right now. If you look closely this is what you will find at the heart of high performance computing. We are turning it into a general purpose computer.

What sets Forward! apart from other similar solutions in the market?
What makes Forward! different is a set of technologies called s-Par — partitioning as opposed to virtualization. The thing that brings the party for us is that if you look at traditional virtualization, only about half of the world is virtualized. So you have to ask yourself "why is that?"

The reason is that workloads are different. It's those workloads that are easy to virtualize that have been virtualized — for instance the web facing tier of an app. So the easy ones are done and the ones remaining are those that have really complicated requirements for deterministic performance; those that can't suffer from any variability in the performance that comes with virtualization. Or it could be that they have a set of security constraints they are just not comfortable sharing a memory space with another operating environment, or that they are so large that it doesn't make sense to virtualize.

There are some who will say that virtualized environments don't give all the business critical attributes that they want, so Forward! is very much meant to speak to this space. We are not trying to compete with traditional virtualization and in fact the architecture says that it runs side by side. They should be orchestrated together.


Previous Page  1  2  3  4  5  6  Next Page 

Sign up for MIS Asia eNewsletters.