AtMos(t)
by Michael Hay on Nov 13, 2008
Okay in the past I’ve blogged about Maui/Hulk in the past and several of us at Hitachi will continue to do so in the blog sphere starting now. The biggest point that I would like to make regarding EMC’s announcement is that they are not the trail blazers they claim to be. That trophy, if there is such a thing, belongs to both Amazon and Google. Some quick math on the lifetime of Google, Amazon, Salesforce.com, etc. shows that these companies have been working on these problems in some form or another on average for at least a decade. It is they who all of us storage companies are following. Further there are other products and other companies that deserve being mentioned, of course both HCAP and HDDS can and do already today act as “cloud infrastructure”. Some of the other companies include Nirvanix, 3Tera and Parascale let alone big G and big A. So to be honest I don’t get why there is all of this hype about innovation from EMC as the first in this space. I suppose again, when EMC chooses to copy the very best they choose yet again to copy Amazon, Google, Parascale, Nirvanix, and Hitachi. Ah now I understand the true meaning of COS it is Copy Other Systems.
Let’s recap a bit the main points that EMC touts as their “innovative” offering:
1. Peta-scale object system with modern (REST & SOAP) and legacy (CIFS & NFS) protocol access with the implication of infinite scaling
2. Single point of management for multiple AtMost(t) systems
3. Policy driven framework for generation of clones, data reduction, versions, retention, pruning, etc.
4. Multi-tenancy to support multiple application sets concurrently
5. Self healing infrastructure which is tolerant of node, drive and other similar failures
I’m sure that there are more virtues or facets to AtMos(t), but this should be a fair representation. Can many of the other offerings from Hitachi and other companies in the Web 2.0/Enterprise 2.0 business do this? The emphatic answer is yes! Sure there are differences such as HCAP (Hitachi Content Archive Platform) supports WebDAV and AtMos(t) does not. Or Parascale, HCAP and AtMos(t) support legacy protocols (NFS & CIFS) and Amazon S3 does not. The important point is that all of these offerings are viable for the “cloud” in some form or another and all sport some cloud optimization in some form or the other.
I know that we haven’t done the flashy marketing blitz to date, but yes I can say that we have real customers who are using HCAP and HDDS (Hitachi Data Discovery Suite). One customer is Swiss Picture bank and I’m proud to say that their site is powered by HCAP (http://www.swisspicturebank.com). As for our other customers, whose names must remain anonymous, HCAP and HDDS are used within both Web 2.0 and Enterprise 2.0 workloads. Just like in a recent post that I’ve done, Hitachi again is the quiet achiever here as in so many things. Past EMC’s marketing hype yes I agree their product is tuned for this market, but I will say this to the folks over at EMC: welcome to the party it took you long enough.
There will be more to come…
Previous posts on this topic:
- http://blogs.hds.com/technomusings/?p=6
- http://blogs.hds.com/technomusings/?p=7
Comments (4 )
Hu Yoshida on 18 Nov 2008 at 6:41 am
Thanks for putting this in perspective.
All this talk about cloud and Atmos reminds me of WideSky, the EMC initiative to control storage management for all vendor storage back in 2001 – 2002. They even had something called AutoIS.
EMC is great at creating visions. Lets see if they can execute better with Cloud and Atmos than they did with WideSky and AutoIS
mhay on 19 Nov 2008 at 1:04 am
Thanks Hu. I agree and I’m a little concerned that unless EMC announces something it doesn’t exist. I think that you point out again that EMC’s marketing is ahead of their engineering. I think that another instance is ILM, what ever happened to that?
Customer Storage Expert on 22 Nov 2008 at 9:13 pm
All the hype about Cloud Computing? Why would anyone want to be associated with something that is essentially just a grid?
The problem I see with Cloud Computing in general is you can’t mix a service grid and a compute grid when it comes to storage. In six months cloud computing will be dead and the industry will be on to the next thing – like SOA.
mhay on 24 Nov 2008 at 3:48 am
Customer Storage Expert, thanks for your feedback. Everyone is entitled to their own opinion. While the time horizon is a valid debate I think that the reality is that there will be a cloud in some form much like there is a power grid. Debates on semantics are always interesting, but eventuality is not deniable. If you haven’t checked out the “Big Switch” I really suggest that you read it.
Also I disagree on the point related to service and storage being mixed in the same physical/logical grid. The whole notion of server containerization or virtualization changes that dynamic radically. Take a look at the XenoServers project and you should get the idea. Storage is really nothing more than a logical set of services which are about persistence of either blocks or files or a mix.
