Enginuity is at the heart of the storage platform that powers more mission-critical apps than anyone else. Enginuity 5875 adds more than I could believe when I first started to look at it many moons ago, and according to Symm engineering is one of the biggest Symm releases ever.
Imagine:
- If I told you “I’m going to give you a software release would DOUBLE your performance for bandwidth-centric, large block workloads!” That’s pretty killer, eh?
- What if that release added the best, most scalable, the highest performance, and the most granular fully-automated storage tiering model… and that capability meant you could do 40% more work, for a cost that is 40% lower?
- What if it delivered a KILLER VAAI implementation – hey, Merry Christmas! EMC now has VAAI support in the midrange and the high end enterprise arrays. I **think** this makes us the only vendor with VAAI support across the midrange and the Enterprise storage market.
- What if it added scalable data at rest encryption and embedded key management – and that capability had ZERO impact on performance?
- What if that same release enabled something that – as far as I know is a completely new innovation in storage land – the ability to federate storage from one array to another – live, without sticking something in the middle? I know my readers are more VMware-centric often than storage-centric. The analogy is that you can “storage vmotion” from one array to another, but in a way that is much faster, and can operate across many different host types. It can do this without adding any “virtualizer” in the front.
That last one (Federated Live Migration = FLM) is an interesting one. If you think about it, the storage protocols that exist are all about a host talking to an array and vice-versa. It’s weird that no one thought to add to the standards the ability for an array to talk to another. That’s the core innovation involved. While replication of the data between arrays has always been there, the final cut-over was always the tricky bit, and involved downtime.
FLM has one array communicate with another and take over the “persona” (all the identifiers) of a storage device. Pretty darn cool :-)
While at first it’s for DMX to VMAX (and VMAX to VMAX), the idea of storage federation will become a universal EMC thing over time. We’re also taking the idea to the standards bodies to try to see if we could get the other storage vendors to make it something that the industry as a whole picks up on (which would be really cool).
oh hey, we threw in a nice bonus – it will do a zero reclaim as part of the process, saving you a pile of $$$ :-)
That list of 5 items is a MASSIVE release. That – plus a TON of other features (seriously!) is what was released.
In my personal opinion, if VNXe is the coolest thing we’re launching today, the VMAX updates are going to have our existing customers dancing in the street the most.
Oh, and Brian – that bit in the mega-launch with the MAX 2011 was… priceless :-)
Later, I’ll also post details and behind the scenes on the mega-virtual storage jump – that was AWESOME :-)
There are some great blog posts that cover this immense release in more detail.
Barry Burke has a post that covers the items above (and many, many other awesome features) in detail here: http://thestorageanarchist.typepad.com/weblog/2011/01/3017-vmax-2011-edition-powerful-trusted-smartest.html
Chuck Hollis did a post on the topic here: http://chucksblog.emc.com/chucks_blog/2011/01/symmetrix-vmax-gets-even-smarter.html
The always awesome Itzik Reich did a great one here: http://itzikr.wordpress.com/2011/01/18/emc-symmetrix-vmax-enginuity-5875-fast-vp-vaai-making-the-best-array-even-better
Would love feedback!
I'm glad to see large block workloads are being improved (I was this customer: http://virtualgeek.typepad.com/virtual_geek/2010/02/solving-a-weird-slow-performance-cloning-issue.html)
To be fair though, #5 isn't a new innovation in storage land. Compellent has been doing it for a while now. They call it "Live Volume".
Posted by: Rick | January 18, 2011 at 11:27 AM
"EMC now has VAAI support in the midrange and the high end enterprise arrays. I **think** this makes us the only vendor with VAAI support across the midrange and the Enterprise storage market".
That would be incorrect. NetApp has VAAI support in Data Ontap 8.0.1 for FC, FCoE and iSCSI datastores. Since all entry level, midrange and enterprise storage controllers all run the same Ontap, that means VAAI support throughout - but you already knew this ;-)
Posted by: Ian | January 23, 2011 at 04:24 PM
Ian, I disagree.
It's not my definition, but in general the "Enterprise" array segment is characterized by arrays that have many processors, large shared, global caches, and support a very broad host-attach profile (including mainframes). In this category, you've got things like EMC VMAX, HDS VSP, IBM DS8000.
In my experience, when a customer's requirements put them into that class of platform, generally we don't see them looking for VNX, and we don't generally compete with NetApp. Sometimes they put Celerra Gateways, or v-Series NetApp platforms in FRONT of those arrays to support NAS use, but that's a different thing altogether.
That's not to say that the the larger FAS6200/6000 (and other FAS arrays) don't serve enterprises, just like the VNX series does (and their smaller brethren).
But - those segments, and the architectures that support them are markedly different. I'm not making a qualitative judgement about good/bad - that's up to customers. Those customers pick those "Enterprise" (I wish there was a better word) architectures for many reasons.
Does that make sense?
I'm curious - do you find many customers transitioning to 8.0.1 in 7-mode (or cluster mode for that matter?)
Posted by: Chad Sakac | January 23, 2011 at 04:55 PM