Monday, November 10, 2008

Defining Cloud Optimized Storage

With today's EMC announcement of their cloud optimized storage (COS) platform called Atmos, we are starting to see the first time an enterprise ready attempt at a global cloud storage system. For the most part, these types of global distributed file systems have been what Chuck Hollis at EMC described as home grown solutions built by academics or hobbyists.

Personally, what I found even more interesting then the actual product release was in how they described a new cloud optimized storage market segment.

EMC describes cloud optimized storage as "the ability to access applications and information from a third-party provider—like a large telecommunications company—that has built a global cloud infrastructure. That cloud infrastructure will make massive amounts of unstructured information available on the Web, and will require policy to efficiently disperse the information worldwide."

One of the biggest limitations to the adoption of Atmos is that it isn't open source. Cloud computing is about ubiquity. The more users of your platform the better. I think that ultimately EMC's activity in the cloud storage sector will help drive more interested and demand for cloud storage across the board. I believe that the rising tide floats all boats and my boat has already left the harbor.

To give you some background a while back I came up with a term I described as the "Content Delivery Cloud" I think the approach of EMC's cloud optimized storage fits into this concept very nice.

In partnership with PandoNetworks, we created a joint site promoting this concept at

Here is an overview:

A Content Delivery Cloud is a system of computers networked together across the Internet that are orchestrated transparently to deliver content to end users, most often for the purposes of improving performance, scalability and cost efficiency. Extending the model of a traditional Content Delivery Network, a Content Delivery Cloud may utilize the resources of end-user computers ("the cloud") to assist in the delivery of content.


  1. Utilizes the unused collective bandwidth of the audience. Every content consumer becomes a server, offloading bandwidth demand from central CDN servers, thus cutting bandwidth costs and boosting media monetization margins.
  2. Improves delivery performance by providing data from a virtually unlimited number of servers in parallel.
  3. Scales with demand. The more consumers demand a particular piece of content, the larger, better performing and more cost efficient that content's delivery cloud becomes.
  4. Benefits all participants in content delivery value chain. To be successful, a Content Delivery Cloud must provide value for the content publisher, the Content Delivery Network, the Internet Service Provider, and the content consumer.
  5. Utilizes a wide range of delivery strategies. Maximize performance and economics by optimally utilizing all available, appropriate data sources, including origin servers, CDN servers, streaming servers, cache servers, and peers. Participants in the delivery cloud can include not only desktop computers but also set top boxes, file servers, mobile devices, and any other Internet enabled device that produces or consumes content.
I also wanted to follow up from my previous post on Atmos.

storagezilla said the following.
Atmos is *not* a clustered FS nor does EMC see it as a Isilon or OnTap GX clone/competitor.

Chuck Hollis, VP -- Global Marketing CTO EMC also had some interesting insights.

The traditional storage taxonomy doesn't do a good job of describing what Atmos (and, presumably, future solutions from other vendors) actually does. As you'll see shortly, it isn't SAN, NAS or even CAS. So, what makes "cloud optimized storage" so different? The use of policy to drive geographical data placement.

He goes on to give some more techical details
Is Atmos hardware-agnostic? Yes, that's the design. It runs well as a VMware guest, for example. That being said, our experience with customers so far indicates a strong desire for hardware that's built for purpose -- especially at this sort of scale. Check out the rest of Chucks post to gain a better insight into the usage and deployment of Atmos.

Here are a few more Atmos links as well:

Steve Todd's,
Chris Mellor at The Register
Network World
Tarry Singh

#DigitalNibbles Podcast Sponsored by Intel

If you would like to be a guest on the show, please get in touch.