ORIGINALLY POSTED 30th March 2010
14,050 views on developerworks
I wrote this a few weeks back – was waiting for the right time to post – and for once it seems Hu and I are coming from the same place!
Its likely a lot of my regulars have heard of Yotta Yotta. To all extents and purposes they produced and interesting SAN cache prototype. It used lazy writes to enable a local 2-way cache to be more dispersed but lazily updating writes to remote caching nodes. It was a kind of competitor to SVC, but it never really got any market acceptance, and the interop was non-existent.
When any company goes into receivership, there is usually a good reason. Sometimes its just the wrong time or place, but more often its due to some fundamental flaw in the product or business strategy
I was interested when EMC bought up the IP after YY went awol, Those following true storage virtualization in the way that SVC provides will understand that EMC backed the wrong horse. Invista… where is it now… does anyone actually produce intelligent switch blades now? I’ve not seen or heard of any that operate at 8Gbit. Not only that, but basing your SAN intelligence on some outboard Windows boxes doesn’t really fly. It may work for an over-hyped mid-range box like Clarrion – it still amuses me, that sticker covering the first four drives…
Out-of-band storage virtualization is dead. Invista, dead. Even the only company that had found a way to actually make it work, Incipient, they too went into receivership last year. So it may come as no suprise that EMC have realised their mistake, and now have their in-band virtualization product about to hit the streets. Some EIGHT years late, based on yet another failed venture, YY.
The YY caching does provide the ability to tie in with VMware’s Vmotion to move a server and storage image from one site to another, but how many people actually need to do that on a day to day basis. While its a “neat” feature… how useful will that be to you today. Do you need to move a couple of server images from one datacenter to another, cool, well that will work for you…It will be interesting how EMC spin this one, and if customers trust them after the disaster that was Invista – which only a few years ago they were telling us was the answer…
What interested me even more was to hear from a reliable source that an EMC sales rep claimed the actual reason for this product even existing was to kill the revenue stream that SVC generates for IBM… (Evil Machine Corp indeed)
Now we could be getting there. EMC don’t like SVC. They never have. They never will. Why you may ask. Its simple. It stops them romping in the cash cow that was DMX, and now – if you let them – V-max. The value of SVC is in the software. For years, you may have been paying OTT costs for multi-pathing, the copy services and replication. SVC breaks those bonds. It lets you use no-additional-cost multi-pathing infront of your EMC boxes. It lets you realize that you don’t need that DMX, or that V-max, and what you need is a midrange box. Down-tier, by using IBM’s STAR or DART offerings to show you just how much Tier0 or 1 you really need, and seemlessly migrate away from it.
SVC has been shipping for 7 years, we have over 9 years of interop testing and support.
EMC’s new YY based appliance may well be sold as the answer to your dreams, other rumours (not sure I believe) I’ve heard suggest they may not charge licensing based on capacity, it may well be a one shot charge – and if that is the case, and it’s any good, I suggest EVERY EMC customer stick it over the top of their DMX, V-max or Clarrion and no longer have to pay any license at all… ???
Personally, it would make more sense to add this “appliance/software” to their V-max and offer migration and virtualization much like USP-V, and IBM does with SVC+DS8000, but thats a HUGE interop matrix to catch up on, unless they stick with the vendor lock-in and only offer it on EMC storage!
So lets see how the masters of spin, or as I recently remarked, the Evangelistic Marketing Campaigners twist this one…
When it comes to a truly open, heterogeneous virtualizaton appliance with proven heritage and credentials (SVC) vs, the latest venture coming from the company that vowed Invista was the answer, using the IP from a failed competitor… I know who I’d choose…