Gartner's unveiling of all-flash array rankings in its Magic Quadrant – expected in July this year – has been eagerly anticipated. Which supplier is in which box? Who is included? Now sources who've seen a possible version of the flash MQ have shared their insight with El Reg – revealing some suppliers are in and others ( …
I find it a bit...
...silly not to include the 3PAR 7450 and Compellent SLC/MLC AFA options simply because they are also sold with spinning disks in other guises. HP and Dell have dumped cash into optimizing those platforms for all-flash and they are marketed as being all flash options. So is the VNX-F 7600, come to think of it. They also offer a bit of a value proposition for HP/Dell/EMC existing customers around training, and many of the advanced array features some of the flash vendors are missing are already available to these models which are built on a known base product. I've used them all in their spinning disk forms and they're not lacking for features or performance capabilities.
I think they offer a ton of value in the AFA market, even if they are just flash-optimized versions of rust-spinners.
Re: I find it a bit...
Minor clarification - HP went through some pains specifically to NOT sell 3PAR 7450 with disks. You cannot have spinning rust on that system, though that is purely a sales thing not a technical limitation. They did it explicitly for situations like this where someone could not claim it was not an "all flash" system.
I wish they would allow disks, the extra data cache(double the 7400) would benefit any workload, also faster CPUs to boot.
Re: I find it a bit...
Though, the traditional arrays that add flash, or are all flash, are at a huge disadvantage against native AFA's. Usually lacking in say for example inline dedupe and compression, that most AFA's already have natively.
Most of the traditional arrays that can be sold as an AFA config, are for certain use cases, that often are not the same as a native AFA.
That said, I think if it were to include them, they would be ranked way low, or bring down the overall ranking.
Re: I find it a bit...
AFAIK MOST of the AFA don't have INLINE dedupe and/or compression. Most of them are just JBOF (Just A Bunch of Flashdrives) with no advanced features at all. That's why, IMHO, Hybrid Arrays with Flash drives are still considered a good compromise in terms of performance and RAS features.
[WARNING: PIMPING] It's ironic that HDS was excluded since we are one of the leaders in shipping flash - http://www.theregister.co.uk/2014/05/05/nand_shipper_numero_uno_aint_who_you_think/
Pure and Tegile and the two best platforms. Both do inline dedupe and compression. Tegile is great because it is truly multi-protocol and smoking fast. Pure on the other really know their shit with regarding to writing to disk. They REALLY know flash. To add to that a guy 1 week out of Devry could manage it. Not sure they will ever do true multi-protocol, but otherwise it is a great box.
SolidFire is really nice as well for the SP community.
Pure switch off inline dedupe when the box gets busy. And why are all these afa's so small? I know they claim usable capacity as being after dedupe but even then they top out about 100tb. Why not just make a bigger one. Makes you think that there is some sort of scalability issue they are gradually working out.
Why AFAs are small
Yes Pure dials back their dedupe (they don't really turn it off but limit how deep they search the hash tables) when the system is heavily loaded. Why care. They post-process it later. The only drawback is if you have a system that's 90+% full and 90+% loaded in terms of performance. In that edge case you could run out of space as the system isn't deduping efficiently. Note that edge case is also really bad planning having the system full and busy for an extended period.
Today's AFAs are relatively small for two reasons:
1 - Almost all of these systems use a pair of 2-socket Xeon servers as the controllers. Processing a million IOPS is the limit of the controller. A vendor could hang another shelf of SSDs on the system but that would add capacity without performance which the AFA customer doesn't want.
2 - Limited customer demand. At the current cost of AFA capacity there just aren't that many users that need PB of storage and are willing to pay that price. Most data is cold enough to store on disk and that is of course a lot cheaper.
Note: Pure is a client.
Right on SMT789
This Tegilian loves your comment! Thanks for the post.
Re: Why AFAs are small
If you're spending money on an AFA then you do so for a reason and you want to thrash it ... so it would concern me that doing just that could cause problems!
And in terms of why they are so small, your second point is irrelevant as you're first point says that they can't currently scale due to technical limitations.
And as for your second comment, the AFA's are saying that flash is cheaper than disk so why should that only apply to smaller requirements? If it's cheaper, it's cheaper ... surely economies of scale only enhance that for bigger arrays? What you're also saying is that hybrid arrays are the way forward ... flash where it is needed, spinny stuff when it is needed.
So in defending Pure what you've actually said is that its' architecture won't scale, and flash is actually more expensive than disk despite their claims.
And whilst I'm at it ... what is an 'AFA customer'? There is no such thing. Find me any customer that doesn't have a variety of performance requirements across their environment. There is no such thing as an AFA customer. There are customers who have a performance requirement for some of their data sets but those same customers will have cold data sets so are they a 'spinny disk customer' too?
I really hope that Gartner does a deep dive on each platform's technology and feature set, to keep them honest. For example, if a particular AFA vendor says they offer a 48TB array with dedupe that can do 385TB, they need to show it in action. If they say they have compression, they need to show it being turned on and off. If they say they support the VMware VAAI primitives, they need to check with VMware and see if those primitives are on the VMware HCL. If they say they have patents pending, they need to show the patent applications AND that they are in the name of the company, and not some OEM partner.
Lots of nonsense out there, let's make sure the industry doesn't suffer by having a bad apple ruin it for everyone else.
Don't look to MQs for a technical deep dive. In conjunction with the MQ, Gartner should be publishing an AFA Critical Capabilities study. There you should see more product detail.
Pointless exercise in inventing an unnecessary market segment
Vendor X has admitted that their mainstream storage architecture can't cope with flash and so they acquire an incompatible startup and try and flog you yet another silo... Instead of looking at whether this is sustainable Gartner tries to push another market segment...
It is utterly pointless penalizing integrated architectures by only surveying systems that can ONLY do flash. Flash is no longer niche - within a few years it will start to kill off performance drives so people need to see credible strategies for flash + capacity drives in an integrated infrastructure. These niche flash only platforms will become less and less relevant as flash becomes more and more mainstream.
- Product round-up Ten excellent FREE PC apps to brighten your Windows
- Hi-torque tank engines: EXTREME car hacking with The Register
- Review What's MISSING on Amazon Fire Phone... and why it WON'T set the world alight
- Chromecast video on UK, Euro TVs hertz so badly it makes us judder – but Google 'won't fix'
- Yes, UK. Johnny Foreigner has better mobe services than you