So, over on his blog , Chuck Hollis is slinging
fud mud at the competition and has even resorted to calling them names.
According to Chuck, if you don't do tiering in the box a'la Symmetrix DMX, then you're being labelled as Frankenstorage. Hmmmmm.
It seems obvious that Chuck doesn't like the way that the competition is virtualising lower tier storage arrays behind intelligent controllers, and he's throwing around a little FUD hoping that some of it sticks to the folks who read his blog.
Now, as Chuck is a man of significant influence in the storage world, there is the worrying possibility that some people may take everything he says as gospel! So in the spirit of clearing up some misrepresentations I thought I'd make a comment or two in response to Chuck, relative to the way Hitachi storage virtualises lower tier arrays behind a USP/XP.... . After all, even some
EMC bloggers bloggers who work at EMC extol the virtue of clarifying misrepresentations .
Powerpoints and press-releases
First up, Chuck states that these "Frankenstorage" creations are "brought to life by the magic of powerpoints and press releases". A cheap shot and a case of firing blanks if I ever saw one. The truth is, so far as the USP is concerned, it's actually brought together by a state of the art intelligent controller, and is running 24x7x365 the world over. No magic there, just intelligent controllers. May be Im missing Chucks point on this one?
Stop gap solution
Then Chuck suggests that HDS only brought their virtualisation to the table because at the time they hadn't figured out how to bungle low-cost-low-performance-less-understood SATA into their expensive-high-performance-mission-critical USP.
While there may possibly be some truth to this (I base this possibility on the verify-after-write technique employed by Hitachi for SATA installed to their mid tier arrays) this comment is at best a bit of a stretch and at worst unadulterated FUD.
In my honest opinion, adding SATA in to the USP went against many deeply rooted principles that the USP stood for, and these opinions and ideals took time t overcome. After all, the USP is high performance enterprise class storage, so throwing in cheap low performance components in to it was never going to go down easily or overnight. At least not at Hitachi....
Test out SATA (physical qualities as well as behaviour of the protocol under all circumstances....) in your mid-tier array before supporting it in your flagship product. Seems like a responsible approach to me actually.
In fact, I know of people at HDS who still don' like the idea of housing SATA in the USP. And I personally, still have my reservations in relation to this. I still dont like the idea.
On a side note, my own reservations will be even more apparent with the addition of SSD/EFD to the USP-V on the horizon, and I'll be posting on this in the near future. And yes I admit that Hitachi are sadly behind EMC with regards to enterprise flash drives.
I for one like the idea of hanging crappy SATA off the back of a USP rather than housing it internally. At least this way get I keep my internal cache, disk adapters and other finite internal resources to myself, and don't have to share them clumsy overweight SATA disks. Would you like to share your lunch, including cutlery, with the fat smelly guy from the Unix team? Never mind sharing your home with him!
There are also plenty of customers who are very happy with the HDS virtualisation of lower tier arrays, and are moving full steam ahead planning further roll-outs purchases of the technology as well as integrating it with Dynamic Provisioning (a seemingly good match). These customers genuinely love it and see value in it. Despite the fact that the USP-V now supports SATAI internally.
Nobody else can afford to do it our way
Chuck also suggests that EMC are the only company that could afford to do all the work required to enable everything from F1-race-car-like-EFDs, to articulated-18-wheeler-wagon-like-SATA-II to race around the same track, at the same time. Quite tricky unless you remove the chicanes. Or in other words EMC house the whole family, including aunts, uncles and in-laws, under one roof sharing a single bathroom :-S Personally I'd prefer to have them down the street or at least annexed off at the bottom of the garden.
Anyway, this also strikes me as a bit of a rash comment. The same could be applied to EMCs slow uptake on other technologies. Just form the top of my head RAID6. Do we think that EMC were slow to implement RAID6 because they were too cash strapped to do the "design, integration and qualification work"?
No, rather, I imagine it was that they didn't believe RAID6 was the right way to go with Symmetrix DMX, but eventually came to the party because of customer pressure etc. Funnily enough, the same kind of principles that I imagine Hitachi struggled with over supporting SATA internally within the USP. Two peas from a pod!
The easy way out
Then Chuck suggests that the competition are all taking the cheap and easy option with their virtualisation approaches, and "when the next array redesign comes around in a few years, maybe they can take a look at native support".
Are we to think that being first to the table with a technology is what counts? On the flip side of this, I could quite easily feel that EMC have all too easily abandoned their morals and the core principles behind their pride-and-joy thoroughbred Symm, and were all too quick to prostitute and pollute it with all kinds of previously forbidden technologies.
Moshe Ya The fathers of Symmetrix would no doubt have turned in their graves at the sight of SATA in the Symm. Not under their watch.
The customers view
Chuck then attempts to view the options from a customer perspective rather than a vendor perspective. Easier said than done when you are a big cheese at a vendor. May be easier for me, who doesn't work for a vendor or a partner and sits with customers most days of the week.
He talks about extra layers when it comes to management and performance. While I see where he is coming from here I feel he is trying to make a mountain out of a mole hill.
Most places All customers that I know, who have deployed external storage behind a USP get by very well even when they have no real skill with the storage that hangs behind the USP. Usually all a customer has to do on the external array is cut a LUN from a RAID Group and present it on a front end port to the USP. Job done. Forget its even there, well to a point.
Anything else that Chuck may or may not be suggesting here would, in my honest opinion, probably be spreading FUD! It is absolutely not the same as having to manage two separate arrays on a day-to-day basis.
As for performance. Performance is already somewhat of a dark art....... Its not like externalising an array behind a USP will suddenly be making something that was previously simple into something hard. Oh and customers already have servers, Operating Systems, HBAs, switches, storage and the likes from every vendor under the sun, and get by just fine. I dont see this as a problem.
In fact, I'm fairly certain that EMC don't back away from customers who run HP or IBM servers and say "sorry we cant help you here, an end to end HP or IBM solution would be much better for you when it comes to troubleshooting....... putting our storage in would only add extra layers of complexity and make things messy....."
According to Chuck nobody likes it anyway, apparently
Finally, Chuck doesn't think anyone really likes the externalised storage approach anyway, he thinks they're all forced in to it by their vendors. I quote - "I don't think anyone really prefers this kind of approach -- I think they're sort of forced into it by their vendor....... they hold their nose, and move forward."
Well on that matter, Chuck, you are flat out wrong. Of course not everybody will like it, but the fact is, lots of people do. I see happy customers all them all the time, and I don't work for a vendor or partner so I feel get honest opinions.
So, as Chuck took the liberty of drawing a comparison, let me draw one of my own, albeit tongue in cheek -
HDS USP/HP XP, external storage = Philharmonic Orchestra.
EMC Symmetrix DMX, tiering in the box integrated approach = One man band, where one guy is trying playing all the instruments himself.
DISCLAIMER: I don't work for Hitachi, HDS, HP or anybody even like that. My opinions are my own, as are my interpretations of what Chuck is implying.