DataDirect Networks Inc. is bringing out an object-based scale-out cloud data storage system called the Web Object Scaler (WOS) to challenge EMC Corp.'s Atmos, and vie for the attention of Web 2.0 data centers.
WOS, in beta now and expected to be generally available in the third quarter, is the latest step in DataDirect's strategy to move beyond the niche of high-performance computing (HPC). Like EMC Corp.'s Atmos, Web Object Scaler scales into the petabyte range -- up to 6 PB in a 100-node system. However, the minimum capacity of two 7.2 TB nodes is smaller than the Atmos starting configuration of 120 TB.
Web Object Scaler nodes come in several configurations. The 3U WOS 1600 is available as a 7.2 TB SAS node optimized for performance or a 16 TB SATA node optimized for capacity. The 4U WOS 6000 can be configured for capacity, performance or both, and can hold up to 60 TB.
According to EMC, Atmos can scale to multiple petabytes and billions of files. DataDirect Networks claims it can handle more than 200 billion files and 6 PB, but also said its system can perform faster than Atmos because Web Object Scaler holds object metadata in memory on its server nodes. Atmos metadata is partitioned and stored in a collection of databases spread across many disks in the system.
Josh Goldstein, vice president of product management at DataDirect Networks, claims Web Object Scaler can perform up to 1 million random file read operations per WOS cloud. He said Web Object Scaler can also provide some of the functions of a content delivery network (CDN) with object policies that can specify which geographic zone data should be stored in and delivered from, placing it closer to the end user requesting a file. The system also automatically finds the fastest path to deliver a file, another feature of CDNs.
More on cloud data storage
New data archiving products focus on software-only delivery, cloud integration
Rackspace cloud storage general manager discusses the differences between private, public clouds
TweetPhoto cobbles together cloud services for storage
Jon Martin, director of product management at EMC's Cloud Infrastructure Group, said the company doesn't publish the number of file reads per second that Atmos can achieve. "EMC Atmos is architected for multiple readers/single writer pattern," he wrote to SearchStorage.com in an email. "The max number of reads is implementation specific and can depend on a number of factors such as [the] number of nodes in an EMC Atmos cloud, network latency and topology, file read size, number and location of concurrent readers, and, of course, number of replicas."
Martin also said Atmos isn't intended to act as a content delivery network.
DataDirect Networks also strives for a simpler user interface with Web Object Scaler by offering a software agent that resides on a customer's application servers and translates between user upload/download requests and placement of data on the back-end system. A user's application must still write to the API included in that software agent, which DataDirect Networks calls WOS-LIB. This is similar to the API integration required to tie in with Atmos. However, the management GUI for DDN's system offers a centralized management point, which Jeff Boles, senior analyst and director of validation services at Hopkinton, Mass.-based Taneja Group, said should be an appealing feature.
"I think it will be a little while before performance becomes a bottom-line question with systems like this," Boles said. "It's more about simplified management and ease of scalability."
DataDirect Networks also intends to compete with EMC on price, because customers can start with a smaller configuration to dip a toe into the scale-out system. Once it scales up, according to DDN's Goldstein, pricing will depend on the number of sites and the amount of data stored, but will generally fall around $3 per gigabyte. "They're offering scale-out technology that can drop into the data center with two nodes and remain price competitive with scale-out NAS," said Terri McClure, an analyst at Milford, Mass.-based Enterprise Strategy Group.
'The Last Mile': Cloud data storage market still taking shape
Although Hewlett-Packard Co. has joined EMC in pushing out what it calls a cloud data storage system, those vendors and DataDirect Networks might end up competing mostly with Web 2.0 service providers' internal expertise.
Tony Kapela, special projects director at New York-based IT infrastructure services provider Voxel dot Net Inc., said the decision for a company to go with a storage vendor's prepackaged infrastructure or build its own depends on "the tradeoff between developing custom infrastructure yourself or doing the integration necessary to get a vendor's canned system into your environment."
Kapela said Voxel is evaluating "canned" clustered file system products from scale-out vendors Isilon Systems Inc. and Ibrix Inc. He's also working with object file systems supported by the popular Filesystem In Userspace (FUSE) interface that lets his customers interface with their file data through standard file manipulation methods. He said the biggest challenge is to mesh the operating system into a "completely different paradigm" of logical volume-based user object storage quotas. "This is an obstacle facing most clustered file systems developed before Web 2.0," Kapela said.
EMC's Atmos is used for AT&T's Synaptic Storage as a Service, and several enterprise customers at this year's EMC World said they were interested in exploring the system. However, "the market uptake is going to be slower than everybody thinks it's going to be," Taneja Group's Boles said. "These platforms are fundamentally about a new way to do core storage."