EMC FAST (the storage equivalent of VMware DRS) is GA…

So, to my VMware-centric readers, you may have no idea what I’m talking about.   But if I asked you “don’t you wish Storage did what DRS does?” – well the answer to that would be YES, right?

Fully Automated Storage Tiering (FAST) is the core idea of automatically moving information to the right tier at the right time.

FAST is immediately available accross the EMC platforms.   On V-Max, CX4 it automates the reconfiguration of LUNs to optimize against policy.  In the V-Max case, it will automatically move and swap LUNs between solid state, FC and SATA.  In the CLARiiON case, it will recommend and move from FC to EFD and SATA to optimize the total configuration.    On Celerra it will automatically and transparently move files between filesystems/tiers on a platform, between platforms, and actually to Atmos (whether it’s Atmos in your internal cloud or external cloud).    Through 2010, FAST use cases will continue to expand in breadth and depth.

So – if you’re scratching your head and the impact doesn’t immediately make sense (perhaps you’re not a storage person) – here’s a way to think about it….

• How much more efficient does fully automated DRS make a VMware cluster than static “VM placement” logic?
• Would you view your compute/memory as “one giant pool” without DRS?
• The idea of just powering down a server – it’s crazy, right?   Well, without DRS, DPM is impossible – DRS is what makes it possible.

Think of it this way (making the analogies):

• VMotion ~= Virtual LUNs/Filesystems or more generally “virtualized storage” (non-disruptive manual reconfiguration of a LUN or a filesystem)
• DRS ~= FAST (automated policy-based reconfiguration of a LUN or a filesystem)
• DPM ~= FAST with disk-spin-down (or automated movement to Cloud storage).  Without FAST, the idea of a “spun down” primary storage use case (different as an example from a backup to disk use case) is impossible.

If you want to understand this better – read on….

Here’s the process of configuring a FAST policy, mapped out for the V-Max use case:

It’s also great to watch, which you can do here…

I wish there was more time in the day – because lost in the huge announcement that is the FAST GA announcement are all the other things now available….

• Thin Zero Reclaim on V-Max (huge for our customers!) – where blocks containing zeros can be “thinned” (hello eagerzeroed thick VMs on VI3.5 deployed from clones or templates!)
• Dense Unified configurations (doubling the disk density possible in CLARiiON and Celerra configurations) – up to 30 disks in 3U – almost a PB in a floor-tile.
• Celerra NAS Dedupe enhancements making compress/dedupe apply to large files and active files (and yes, there’s a plugin coming for that).
• Unified configurations now can have all block types with backend connectivity (and wait till you see what’s up our sleeve around management models).

We’re not stopping here.   FAST for most customers is a 30%/30% improvement (30% more performance out of a configuration that is 30% less/cheaper) – even without talking about FAST v2.  Thin Provisioning adds about 40% more efficiency vs. thick – again for most customers.   Primary storage dedupe and compression adds another 40-50%, and in backup use cases can squeeze out even more than in the production use case.   “Green” (low-power and spin-down) can lower power costs out the yin-yang.   “Gone” (take the data right out of the system – either deleting or export to external cloud storage) can save a ton.    The more you can apply these technologies, and the more pervasively you deploy them – the more ($, watts, square feet) you can squeeze out.

We are working around the clock to reduce your storage costs, and make storage management increasingly invisible (big thin pools that auto-tier).  Why?  Does that seem counter-intuitive?  Man, oh man, it’s not.   Talking to a FT10 customer last week, their exact comment to me was:

“if we extrapolate our storage growth over the next 5 years, even assuming a 30% increase in power and cost efficiency every year,  we’re not going to be able to physically HOUSE it in our datacenters”.

To contain this – we’re employing every technology out there. Hyper-Fast SSDs+Huge SATA+FAST (today).  Dedupe/Compression everwhere we can – production and backup (today).   Cloud Storage models that glom DAS and servers into exabyte-scale clouds, and NAS that can automatically leverage that as needed (today).   Fast v2 will be here before you know it also – extending this so it moves sub-LUN elements on block devices.   If only there were a way for that to integrate with VM-level awareness and policy :-)

If you want to learn more about FAST, here are some sources to check out:

The official product pages:

• V-Max FAST
• CLARiiON FAST
• Celerra FAST

EMC blogs:

• Chuck Hollis: FAST and the Continuing Virtualization of Storage (Part 1, Part 2, Part 3)
• Storagezilla: FAST to the FULLEST
• Gina: EMC FAST – how do IT folks keep up to speed?
• StorageAnarchst: EMC FAST and the big 5

Non EMC posts:

• StorageNerve here
• The Register (Chris Mellor – uncannily accurate as always, but always also a characteristically pessimistic :-) here

This association between the idea of FAST and VMware DRS isn’t just a passing analogy.    FAST configuration is already relatively simple right now and occurs at the LUN level on the block devices (and the file level on Celerra) we are actively working to integrate with with Storage DRS (demontrated by VMware at VMworld 2009) with FASTv2 on EMC V-Max and our Unified platforms.

Like I always say – listen carefully to our VMworld keynotes – I ALWAYS (and I will keep doing it) have divulged things (I have to tiptoe a bit to avoid getting in TOO much trouble) that show up with 6-12 months in each of the EMC/VMware (and VMware/Cisco/EMC) keynotes :-)

Watch the VCE one from 2008, and you can see all the UCS setups, as well as what we were doing around 10GbE and vCenter integration.   Watch the VCE one from 2009, and you can see the VCE launch setup.   Watch the VMware/EMC one from 2009, you can see the FAST setup, the vStorage APIs for Array Integration setup, the vStorage DRS setup and in TA3105 the geo-graphically disperse VM teleporter setup :-)   There are hints in the 2009 session of some other crazy stuff targeted for 2011.

Man – I love my job!

Comments

[Len]

great blog, thanks for the info.

[PiroNet]

Can we say FAST is the Compellent counterpart but (much) more cheaper and not as advanced?

I guess the wide availability of the EFD at EMC has leveraged FAST :)

Cheers,
Didier

[za_mkh]

I also thought that this is just the same thing Compellant has been doing for years. Question is .. Is EMC going to be charging for this?

[Chad Sakac]

@pironet, @za_mkh - there's no question that Compellent was the first mover (IMO) around "automated tiering". I certainly don't think we're trying to take away from that. Analagously, Thin Provisioning had 3PAR being the first mover. Datadomain was the first mover for target-based data dedupe. Copan was first mover on dense storage configs. Netapp was first mover on unification of NAS and block. EMC was the first mover on consistency tech (years ago) EFD in enterprise arrays (recently), and the idea of scale-out x86-based enterprise (any IO served by any port/brain/cache/bus/disk + support for open systems and mainframe) platforms.

Each of us innovates in different ways, each of us does something first. When someone does something that is a good idea (fully automated data movement/optimization) - the industry as a whole moves.

I think eventually this idea of leveraging multiple tiers automatically will (likely) become something everyone needs to do.

What we think we do well:

1) we think that the maximum value of the idea of FAST comes from EFD coupled with massive (but very slow) SATA/SAS (we now offer 2TB SATA in our configs)
2) at very large scale - the number of simultaneous moves and speed/impact of the mobility actions becomes really, really important.

These are just some of the things we think people will like about FAST.

Also, of course, there are loads (loads!) of things we do on top of FAST. Think VMware integration (vCenter plugin and vCenter API), production dedupe on NAS (heck, NAS in general). Support for dense configurations (30 disks per 3U) came out (as well as thin zero reclaim on symm).

Will we charge for this? The FAST pricing starts between $5K-22K.