Francis Fukuyama wrote 'The End of History and the Last Man' in which he famously postulated that, as far as systems of government and markets were concerned we had reached the 'end of history' ...democracy and the free market had defeated all comers and, as far as Francis was concerned, the only thing left open to debate was how to implement such systems and what controls were required for regulation. The book became a bit of a touchstone and rallying point for what would become known as 'neo-cons', but as time has marched on many of Francis' original assertions have been challenged by the likes of Robert Kagan in 'The Return of History and the End of Dreams' ...as well as real world events like 11 September, the Iraq war, and the recent economic recession. The end of history? Doubtful. Just as surely as we have debated systems of government and markets since Greek and Roman times, so we will surely debate them in the weeks, months, and years ahead.
Whats this got to do with storage?
I disagree with Joe Tucci ...SSDs aren't going to have anywhere near the impact on the storage industry that he thinks [or hopes!] they will. No, I think the end of history for storage ...or the beginning of the next evolution of storage, if you prefer ...belongs to policy based promotion/demotion of storage as well as policy based data mobility cross product and/or cross vendor.
The emergence of technologies such as EMC's FAST [Fully Automated Storage Tiering] are interesting not just because they could introduce policy based or SLA based promotion/demotion [data placement on SATA, monitor IOPS and promote to SSD if IOPS spike above SLA/policy] but also because these may be the first shots fired in an effort for EMC to introduce what customers have been demanding for years ...the ability to move data from a Symm to CX to Celerra to Centera based upon policy alone. Automatically. Without intrusion, complexity, or downtime for a customer infrastructure.
Reality? Well, we'll have to wait and see and I am loathe to speculate too wildly about what major vendors such as EMC will introduce in the near future.
What I will be watching, however, is how and where the code which enables FAST comes from.
If it is an organic 'in house' project within EMC, it is probable that FAST will be offered free of charge and will be the beginning of EMC linking their storage products together in a way which won't require highly specialised training to operate for the average customer.
If, on the other hand, FAST is a result of an EMC acquisition ...I would watch Precise, for instance given Avner Yoanei was previously at EMC as VP of Symmetrix SW Engineering for Storage Platform Operations Division ...then it is likely that someone within EMC will want to show a return on that investment, so FAST may come at a cost to customers in the first instance. And I would be very wary of FAST's stability/ability in rev 1 given EMC's track record with integrating thrid party code into their storage products.
In any case, storage management is about to get a whole lot more interesting (and simplified!) for the average customer given the promise of grid storage architectures and policy based management.
The end of history indeed.
UPDATE Mon 19 Oct 2009: A great blog post by Chris M Evans discusses how tiering works today 'manually' on EMC architecture and, perhaps more imporantly why customers would demand FAST to be free. The most interesting comment to Chris' post? Mr. Barry Burke, the Storage Anarchist himself states 'I’ll defer the pricing discussion until after GA announcements later this quarter. And the patent topic really isn’t something I can comment upon at this time.'
Makes me wonder if I'm on to something re EMC's intentions for FAST release and how it will be priced.
Subscribe to this comment's feed
written by mpyeager, October 22, 2009
Whilst I found your recent twitter message a bit well ...condescending ... @mpyeager @BasRaayman @chrismevans Methinks U R 2 wrapped up over FAST pricing. If FAST doesn't reduce CapEx *AND* OpEx, it's NG (duh!) I do understand where you are coming from.
I don't think there is any question that, if FAST doesn't reduce CapEx and OpEx it will be no good. That is frankly a bit on the 'frightfully obvious' list at not at issue. What we have been discussing is the pricing and, if I'm honest ...I personally feel that FAST is more a feature and less a product and, hence, should be offered free to customers. The majority of your comments and those of others would lead me to believe that EMC don't view it that way and will be keen to charge customers for this new 'product'. That said, I don't work for EMC ...nor do I own shares ...so what EMC decides to charge or not charge is entirely up to them. I am, however, still entitled to my opinion as a storage and technology professional.
As regards 'beating up IBM and HDS re thin provisioning', I'm somewhat confused. I struggle to understand how having a professional discussion about what EMC may or may not do in future ...or may or may not charge... is 'beating up' on EMC. Sorry, but being inquisitive and presenting a point of view succinctly as one sees them is called debate where I'm from so I do apologise if by my post or subsequent comments you felt EMC were being singled out as this was ...and remains... not the intention.
If I'm honest, I believe that thin provisioning should be included for customers gratis in all storage platforms as it is less a product and more a feature. I note that you left IBM XiV out of your list of vendors and products you feel I should go 'beat up' in an effort to be fair and impartial.
I can already see the comment(s) coming from you and possibly others decrying that IBM XiV isn't an 'enterprise' product, so let me state here that I view it every bit as much an enterprise product as I do EMC Vmax. In that sense, It would be interesting to note that IBM XiV offer not only thin provisioning gratis ...but also zero page reclamation, aysnc and sync replication, and snapshots as these are features and, frankly, not products.
So, I will cast out my challenge back to you ...and HDS ...are you actively considering including zero page reclamation, aysnc and async replication, and snapshots gratis as part of your storage array offerings?
-
- +0
-
-
written by Bas Raayman, October 23, 2009
First off, you are in the business of making money with (among others) storage and software. You released a vision or a direction and are moving toward said vision. One of the problems with talking about a vision is that people will always ask questions about what you are doing, and more importantly on what you are intending to do.
The fact that you reduce CapEx and OpEx is a no brainer. But it's also not the most important feature. Hell, I would also buy something that actually increases my CapEx and OpEx, as long as said product or solution if it helps me meet or exceed my targets.
Combine that with the fact that a lot of companies don't just look for CapEx and OpEx anymore. I spoke to a lot of colleagues and everywhere you get to hear the same bullsh*t bingo played in terms of quick wins and low hanging fruits. The feedback I got was that a lot of people might like the ROI that a product can offer, but still didn't purchase the same because they simply thought the initial invest was too high.
As David Vellante already stated on twitter, I am also confident that as a feature or as a product FAST v1 and v2 will receive high demand from customers.
In regards to the fact that we "are too wrapped up over FAST pricing", well why shouldn't we be? Any DMX or V-Max will cost you big bucks. Usually enterprises that purchase them don't have to pay the list price, but products in this class of storage will set you back a bundle. And sure, you get a lot of features for the money you pay, but the real question in my opinion is if I want to pay for the extra functionality seperately, or if I can expect this functionality to already be in the base product price.
No matter what vendor I purchase from, I am going to pay for the functionality by either paying it in the base price (and then they will say I get it for free), or I pay for the functionality I use. One could say that with EMC it's not unusual to pay seperately for additional functionality, but depending on how good you know them your mileage may vary.
Oh and Barry, we can beat up on several people, but you guys are currently in the spotlight the most, so I guess we started there. On the other hand most people know you've got a good pair of boxing gloves so no worries there.
-
- +1
-
-
written by chrismevans, October 23, 2009
I don't think anyone has an issue with companies making money. The issue is the mixed message. We're being told FAST can't be free - it takes money to develop: Mark Twomey commented to me last night on Twitter:
@chrismevans We can't raise the capital to buy the rest of the IT industry one company at a time if you don't pay. Uncle Larry gets this!
However we also have the conflicting EMC message that thin provisioning has been given away for free (Barry's comment above). Surely TP took a serious amount of effort to integrate into Enginuity and therefore should be chargeable?
Wait a minute though, we forget market forces. The competition nipping at EMC's heels already offer thin provisioning. Some of them offer it for free. In addition, I suspect that thin provisioning take-up hasn't been as pervasive as EMC would like, because the net saving of TP minus the licence cost isn't that great and doesn't justify the Opex investment to make it work. EMC have made it free to increase adoption and match the competition.
So in reality nothing changes. EMC and others will choose to give away their products where they feel they are losing market share and need to gain it back or it gives them another competitive edge to do so. Hitachi have done the same thing with "Switch it On" http://www.hds.com/corporate/p...90422.html
As Bas says, why shouldn't we be concerned with what things cost - unless of course you are Larry and can afford to pay, regardless of the cost.
-
- +0
-
-
written by storageanarchy, October 23, 2009
Seriously, guys, all this ruckus, challenges and accusations, even before you see how it is priced/packaged...
And all this negativity. WOW!
Let's pick this up after the pricing and packaging is announced. You just might be pleasntly surprised...
-
- +0
-
-
written by BasRaayman, October 23, 2009
On the other hand, I would be much more interested in the technologies involved. Either way we will need to wait for the final pricing to come out. And who know, perhaps this 3 on 1 might even help set a price point or range when EMC is considering the options.
So cheer up Barry, we still like you and the tech you bring with you.
-
- +0
-
-
written by mpyeager, October 23, 2009
*
Yours, and EMC's job is ...frankly ...to corner the market on storage and sell storage everywhere in every iteration possible (i.e. enterprise, modular, BURA, CAS ...you get the point I'm sure). I don't fault you or other EMCers for taking a 'whatever the query is, the answer is EMC'. No one, least of all me, would ever claim that one should be denied a livelihood.
*
That said, my job is to interpret and validate what vendors make ...and say ...in the pursuit of ensuring that what I as a Practice Leader and technologist recommend 'stacks up' both technically and in support of demonstrably reduced OpEx and CapEx costs not just now but, preferably for a three to five year cost benefit analysis period or better. This is the value that I continually endeavour to bring to my customers.
*
I make these enquiries and validations for all of the vendors I work with ...EMC, NetApp, HDS, IBM, HP and several 'Tier Two' vendors.
*
If you felt that we were 'beating up on you' well, sorry for that ...but you need to protect the EMC patch and I am entitled to my opinion. Particularly when those enquiries and opinions are in the interests of providing value to my customers, and in the interests of being fair and equitable to all of the vendors I work with.
*
I don't accept that we were being negative, but will hold you to picking this up when the pricing and packaging is announced. Here's to me (and others) hopefully being pleasantly 'surprised'.
*
/endthread : /pull pint > disti @storageanarchy, @baasryman, @chrismevans
-
- +0
-
-
written by storageanarchy, October 23, 2009
I've also provided a pretty detailed description of what FAST is, and isn't, for both V1 and V2 over on my blog - (http://thestorageanarchist.com)...there is no need for speculating about what it might (and might not) do...I think I've spelled it out pretty clearly already. If not, I'll be happy to answer any questions I can about the functions of FAST, feel free to ask questions on my blog.
And while indeed there will be a price assoiciated with FAST, the net benefit will be reduced CapEx, OpEx and TCO...and indeed, storage management is ALREADY getting a whole lot simpler.
-
- +0
-
-
As to the topic of price, if FAST is to be at all successful, it not only has to automate placement, deliver predictable performance and simplify storage/performance management, it must *also* reduce CapEx, OpEx and TCO vs. non-FAST, non-Tiered or hand-optimized tiering alternatives. It's that simple.
IMHO, of course ;o)
May I also remind you and your readers that EMC is the ONLY enterprise storage platform provider delivering Thin Provisioning for free. That's right - the software that enables you to use LESS Symmetrix storage, to IMPROVE your capacity UTILIZATION, is actually $0.00 FREE for Symmetrix DMX and V-Max. No license fee, no maintenance fee.
FREE.
Perhaps you might beat up on IBM DS8K or Hitachi/HDS/HP/Sun USP-V for charging extra for their Thin products? Their products are GA; they have nothing to hide behind B^]