New performance planning considerations for V-Max

Geoff Hough,  Sr. Director, Business Strategy

Students in EMC’s 4-day class for understanding and planning Symmetrix performance may need a new module to cover what’s changed with the new V-Max, just as they have today to cover the differences between  Sym 8000 and Sym DMX arrays.

https://education.emc.com/content/_common/docs/training/Customer%20Catalog.pdf  (page 19)

For the course designers though, the good news is that they might be able to borrow some of the concepts from the 3-day Clariion performance planning class.

 

https://education.emc.com/content/_common/docs/training/Customer%20Catalog.pdf  (page 17)

The reason is that with the new V-Max architecture, CPU/bandwidth resources are now shared between frontend processes and backend processes.   Clariion users are used to this, but users of legacy Sym architectures are not.  They are used to having dedicated resources for frontend processes and to balancing host access across these resources without much concern for backend disk management processes.  Likewise,  they were able to plan and balance backend processes across dedicated backend resources without much concern for frontend patterns and resource utilization. The differences between the old and new architectures are illustrated below.

Time will tell how users and performance consultants react to this.  One supposition is that users will now plan to assign and support applications on an engine-by-engine basis in order to avoid reconciling FE/BE resource utilization among multiple applications. If so, individual engines could be planned and configured using principles similar to those used with Clariions today -- a boon perhaps for IT shops that will be running these two different platforms? 

Comments

[the storage anarchist]

You might have a point here, if you actually understood the V-Max architecture.

But you clearly don't.

Being as the "Old" vs "New" diagrams above are probably from 3PAR's competitive response deck, I won't help you out by correcting the inaccuracies and misrepresentations.

Suffice to say that prospects briefed on V-Max will immediately recognize that 3PAR doesn't know what they are talking about when it comes to V-Max.

[Geoff Hough (3PAR)]

After all the terrific marketing content used to launch V-Max (including the numerous, selected technical deep dives) one might reasonably expect to understand something as straightforward and relevant as likely data movement and layout within the machine -- without requiring a special "prospect briefing."

If as many as four days are required to explain these dimensions or subtleties, perhaps that is reason enough. But it begs the question: What other 'unknowns' must wait to be revealed in a private briefing? What will be told as a matter of course and what must be specifically inquired about?

[the storage anarchist]

No private briefing required - come to EMC World next week and see for yourself where you have mis-interpreted the architecture and the implementation.

There are over 56 sessions that will cover various aspects of Symmetrix V-Max, including performance and configuration best practices, along with hands-on demos of the new automated provisioning and simplified management. In fact, I think you'll be quite impressed at how well V-Max implements thin provisioning and wide striping!