I was chatting with fellow ESG'er and avid HDS proponent Tony Asaro about an unrelated matter - how HDS has really been one of the few to take advantage of the V word - they use it as a differentiator on their big Tagma USP gear and it's getting them deals. People aren't buying it just for the virtualization - but with all (perceived) things being equal, folks would rather have the next next thing than not. Makes sense.
Overall, HDS is doing great, except in one area. I can't figure out why they are letting their biggest block competitor, EMC, run unchallenged with their Centera all over the place? EMC is pulling a half a billion a year, directly, and lord knows how much more in services and other products, with Centera. It gives the EMC folks cool - strategic - initiatives that are very topical to discuss inside a big account - and not just "my array is better than their array". The EMC guy gets to elevate their game by talking compliance, risk mitigation, and higher data oriented constructs to non-storage traditionalists like the legal folks, the risk officer, etc. Those guys have way more credibility and access up the ranks, which directly equates to bigger buckets of money and higher strategic visibility.
I know it isn't a technology issue, they have had Archivas software, and lord knows they have plenty of hardware, so it has to be a focus and packaging issue. The fear is by the time they figure that out, EMC is on to the next big thing and HDS ends up chasing the wrong market.
It's not that IBM or HP are doing much better here, but at least they have products folks can buy. Let's get in the game people!
On a side note, the next big opportunity for object based storage is going to happen in application areas where there is enormous quantities of fixed content/persistent/non-changing data that rely on a traditional database to be able to query it. Those databases aren't made to be effective in finding a needle in a haystack, they are built for transactional speed. An app vendor who has to pay an Oracle license that's as big as the revenue they get from a customer would be pretty happy to get rid of that expense, and if you could tell a user they could mine 100X the data in one tenth the time, I'm pretty sure you'd find a whole new market psyched to give you money.
Or, how about a unified location for everything non-transactional? Logical or physical, if everything in your world was on one thing, thru one object based meta-data smart system, then finding things in a federated fashion doesn't seem like so much of a pipe dream.
They've had a product on the market for a while now, Content Archive Platform. Asaro even has written about it. So what are you talking about here?
Posted by: Scott | December 19, 2006 at 03:43 PM
"Those databases aren't made to be effective in finding a needle in a haystack"
I'm sorry, but this is exactly what rdbs theory is all about; look at all those indexing and search techniques that are implemented in modern databases! I suggest that saying that you can easily come up with something new that "could mine 100X the data in one tenth the time" is a bit lame
----OK, i'll play - then why is it that everything isn't stuffed inside a database? It's easy to make unstructured or semi-structured data structured, so why don't we? Why bother with a file system if the dbase can manage the contextual relationships? The answer, my friend, is that the current database methodology that is used in commercial computing is designed to be transactional in nature, so performance is the key criteria. Volume eats performance, so it's only natural that a database with "fast" as it's raison d'etre can't also be huge - which is why the average commercial Oracle database is still measured in GB's, not TB's or PB's. In a perfect world you would have every bit of data in a structured field, because that's the easiest thing to search against, but if you have to scan a billion rows and a zillion columns, performance will still suck. If you don't have the fields intetified up front, you can't look for context.
Don't worry, I think I found the answer, but before I give some little goofy company credit for changing the world I need to do some more homework. Thanks for the note. I'm going to make this discussion its own thread. Cheers
Posted by: Alex | December 20, 2006 at 06:24 AM
Hi Steve, just to let you know HDS has products people can buy. The HDS CAS offering (we call it Acitve Archive) is called HCAP (Hitachi Content Archive Platform) and it's GA since ocotber.
---Well, I guess my point is made! I knew there was something, but clearly if I don't know about it that means you guys aren't pushing it, because users (nor competitors) are talking about it. Time to let the secret out before your friends eat all the pie, as it were. SD
Posted by: Juerg Maier | December 22, 2006 at 03:24 AM