EMC Corp. today made its first update to its Atmos cloud storage system since its release in November 2008, adding...
a series of high-density hardware configurations and a new data protection option called GeoProtect.
The next generation of EMC Atmos hardware, the WS2 series, adds Intel Corp.'s Nehalem quad-core processors to the server component of the rack, and 2 TB SATA drives for double the capacity of the WS1 models. EMC has also added memory to the new hardware configurations, which has the company claiming a 50% total performance improvement over the WS1 series.
The WS2-120 can now offer up to 240 TB capacity, and a half-populated 60 TB version of the WS2-120 is also being made available for entry configurations or remote offices of large enterprise companies. The WS2-240 holds up to 480 TB, while the WS2-360 holds 720 TB.
Atmos 1.0 supported replication of full copies of objects among geographically dispersed systems as the only option for data protection. Atmos 1.3 adds GeoProtect, a RAID-like option that splits fragments of data within each data object among geographically dispersed locations. With GeoProtect, data can be restored if fragments are lost.
The fragments are geographically dispersed in coded segments. Two options are available with this release for GeoProtect: a 9/12 encoding ratio -- nine fragments are dispersed within 12 coded segments -- for RAID-like data protection with 33% storage capacity overhead; and a 10/16 ratio that can tolerate up to six failures at 66% capacity overhead.
John Martin, director of product management at EMC's Cloud Infrastructure Group, said customers can assign GeoProtect and replication to objects according to policy within the same Atmos system. Because there's latency involved in reconstructing an object from geographically dispersed fragments, GeoProtect will likely be used for "content that needs to be retained but infrequently accessed." For lifecycle management, customers can replicate objects for a certain length of time, then store them using GeoProtect for long-term archival retention.
Atmos customer manages compliance regulations
CareCore National has two Atmos deployments at data centers in South Carolina and New York as part of a new initiative to "gain control of unstructured data," according to Bill Moore, the firm's chief technology officer (CTO). Atmos was brought in as a replacement at CareCore for a Clariion disk array with a network-attached storage (NAS) head that the company had outgrown. CareCore's IT staff is migrating 500 TB of unstructured data to Atmos.
Moore said GeoProtect will also fit nicely into "the complex regulatory world we live in," where different files and objects have to be managed according to compliance regulations that vary state to state, and often overlap. Moore said managing objects at scale by policy will ease the management burden, and GeoProtect will allow that long-term retention to take place with less storage overhead than was required by Atmos 1.0.
"Rather than multiple live copies, we can use GeoProtect to bring up files in deep storage, but more efficiently use the online space we've got," Moore said.
EMC said it has shipped 15 PB of Atmos capacity since the product was first launched. While it's primarily positioned for Web 2.0 companies, multimedia and telcos, EMC's Martin said health services companies such as CareCore have also begun to form a significant portion of Atmos customers.
CareCore's Moore said integrating Atmos into the existing environment required programming, scripting and other specialized skills among his staff to link internal applications to the Atmos system, which is accessed using REST APIs rather than traditional file network protocols CIFS and NFS.
Moore said CareCore has also written its own internal migration utility to find unstructured data in islands on its network and automatically move it into the new Atmos system, at a rate of approximately 2 million documents a day. A total of 45 million documents will be in the system by the time the migration is finished. "It's not just something you take out of the box, plug in, and you're ready to use it," he said.
Another healthcare Atmos customer, Boston's Beth Israel Deaconess Medical Center, has had its own unique challenges migrating legacy data into an Atmos system. Beth Israel storage architect Michael Passe says EMC engineers have done quite a bit of work over the last year to help him integrate Atmos with his F5 Acopia ARX file virtualization switch, which in Passe's case requires CIFS integration using the Samba protocol.
Now Passe hopes to see the addition of automated failover among nodes in the Atmos system through the Acopia switch, but acknowledges the way he's trying to deploy the product is not necessarily in line with its target use case. "Atmos totally works in Web 2.0 protocols for files like SOAP and REST," he said. "It's why [EMC] strongly encourage the use of SOAP and REST [with Atmos]."