Where is FAST VP?
by Hu Yoshida on Jan 17, 2011
On December 21, 2010, last month Nigel Poulton posted “VMAX Comes of Age” on his Technical Deep Dive blog. It was a very interesting article that claimed that EMC released their long anticipated sub LUN level tiering version of FAST, which was identified as FAST VP, previously known as FAST 2. Nigel claims that it was released on December 15. Nigel goes into a good level of detail describing the implementation of this new feature and compares EMC’s 360 MB extent size to Hitachi’s 42 MB page size and IBM’s sub LUN extent size of 1GB. Since then Nigel posted a comparison of VMAX vs VSP and compared EMC’s FAST VP and Hitachi’s HDT ( Hitachi Dynamic Tiering).
I provided my response in a comment to Nigel’s post here.
During my research for this response, I found it strange that I could not find any reference to this in any EMC press release or on EMC’s website. When I Google EMC FAST VP, I only find references to Nigel’s post by other bloggers. Some sources say that there was a description of FAST VP on an EMC solutions engineer’s blog which was posted on December 17 but has since been withdrawn. All the EMC bloggers who commented on Nigel’s post avoided any reference to FAST VP.
Has EMC quietly slipped FAST VP for the VMAX out the door? Why the uncharacteristic silence? IBM and Hitachi have already delivered on sub LUN level tiering for enterprise storage but, aside from Nigel’s blog, there doesn’t seem to be any other reference to FAST VP. I believe sub LUN level tiering is a game changing technology which will be table stakes for any enterprise storage vendor.
Can anyone help verify the existence of FAST VP?
Comments (9 )
I’m a bit disappointed that you don’t implicitly trust my blog and don’t see it as the epicentre of all good, reliable, independent, unbiased technical knowledge
Although this is somewhat like a self-certifying certificate….. but as I said on my site, EMC went GA with Enginuity 5875 code on 15th December and PowerLink users should be able to verify this. FAST VP is a fundamental part of 5875.
I know FAST VP was released last december but they didn’t publicized it.
I think this is a marketing strategy, tomorrow they will host a big event and VNX (Clariion/Celerra evolution) isn’t enough to justify all the hype.
I have to agree with Enrico, there may be things that will be disclosed tomorrow related to Enginuity 5875 with further announcements to VAAI, FAST VP, VNX and some sort of VPLEX integration / further additions to VPLEX. Unfortunately there was not a mention about 5875 from Mr. StorageAnarchist on his blog about this addition.
Well we are not too far, lets wait and see what comes tomorrow…
This isn’t the first time that EMC have released a feature that at the outset would seem beneficial but they chose not to publicise it. I wrote about another similar scenario a year ago. http://www.thestoragearchitect.com/2009/12/11/enterprise-computing-has-emc-slipped-zero-block-reclaim-into-v-max/
FASTv2 was released late last year (I wrote about it at the time, commenting that Symmetrix FAST v2 was still not GA). http://www.thestoragearchitect.com/2010/10/03/emc-delays-new-clariion-and-celerra/
As Nigel mentions, it is GA now; it was discussed today on the EMC Breaks Records presentation, but it wasn’t the main feature. I would suggest that as the benefits of FAST VP are mainly seen with SSD drives (and I don’t think that many customers are using them) then it’s not the most important of announcements.
VMAX sub-LUN FAST VP began shipping in December, And wa announced today in Singapoer am NYC (London to,or row). It was announced today along with 40 other new products (19 of them VMAX-related) and 55 total new features/enhancements for current and new VMAX arrays. My blog posts and http://facebook.com/thestorageanarchist links cover the launch and explain why VMAX FAST VP is the Smartest implementation of automated tiering – bar none.
I look forward to your responses to everything we announced today, but especially FAST VP and FLM.
Thanks for asking!
the updated post can be located at:
Can some one clear the picture here, 42MB/360MB and 1 GB sizes pages?
Is that same what is block sizes and big block sizes (applications) will take more blocks in disks and small bocks(application) can waste the storage if smaller then disk block.
Dynamic provisioning is an exciting topic , however is this same as snap shots in backups (I understand uses are different). In this case all FAST or Dynamic provisioning would be not able to understand the data structure, it would be only moving the data across the tiers depend on IO loads.
Can you please tell me, when we Celerra customers can have a v-max with the new code. We just got news that there is no Celerra code supporting 5875 so we cannot upgrade our v-max. Still waiting for features promised…
Hello Ripunjaya, please restate your first question so that I can better respond to it.
On the second part of your comment, your assumption is correct, we do not know the data structure underneath. We are only moving pages based on page accesses during a selectable time frame.