News Stay informed about the latest enterprise technology news and product updates.

Scale Computing develops IBM's GPFS for midmarket scale-out multiprotocol storage

Scale Computing's scale-out multiprotocol storage still has yet to add some features on the enterprise storage checklist, but users say the system's price and expandability swayed them.

Scale Computing Inc., which first entered the enterprise data storage market earlier this year, this week released the second version of its storage system based on IBM Corp.'s General Parallel File System (GPFS) for midmarket and small companies.

Scale, which came out of stealth with its Intelligent Clustered Storage (ICS) system in June, now claims 30 paying customers and 50 channel partners. ICS 2.0 features broader data protection and greater expandability.

ICS takes management features and multiprotocol access developed by Scale and wraps them around IBM's high-performance computing (HPC) proprietary parallel file system, including CIFS NFS, and iSCSI, which can all be run simultaneously. This is accomplished through a protocol translation layer that fronts the system.

ICS 2.0 adds support for up to 31 snapshots, developed with Scale Computing custom code written to IBM's GPFS API to provide block-level snapshot storage. In a future release, 256 snapshots will be supported.

More on multiprotocol storage
Nexenta Systems pushes NexentaStor forward with open storage and ZFS

The University of Arizona consolidates networks with Fibre Channel over Ethernet (FCoE) 

Brocade expands battlefield with Cisco to encompass Data Center Ethernet and FCoE

Dell brands Brocade Fibre Channel, Ethernet, FCoE devices

With support for only 31 snapshots and no native synchronous replication, ICS lags behind enterprise storage systems in data management features but early adopters say the price is right.

Bruce Pina, corporate IT manager for Charlotte, NC-based broadcast company Bahakel Communications, said he brought in Scale Computing's system to replace a Dell/EMC AX100 that he used as a backup storage area network (SAN). He said iSCSI partitioning is still limited in ICS, which led him to switch to CIFS so different systems could share resources more efficiently at the same time.

"The fact that we're using it for backup rather than primary data makes me more comfortable going with a new company," he said.

Bahakel still uses a Dell/EMC Clariion system for primary storage, and archives data to tape for additional redundancy. Pina said he appreciates that future features will be added to ICS at the same price, which is already lower than the products from Dell/EMC and Hewlett-Packard that he considered.

Brian Beck, technical director for the GEO Foundation, which manages charter schools in Indianapolis and Colorado Springs, Colo., began deploying snapshots and replication when they became available to early adopters in October. He says the 12 TB capacity spread over three ICS clusters at three different sites cost at least $14,000 less than the 12 TB systems he evaluated from NetApp Inc., HP/LeftHand Networks and Dell/EqualLogic.

"Money is a huge factor," he said. "We're in education -- every dime we don't spend on technology is a dime we can spend in the classroom."

ICS 2.0 also adds a new node, the 4 TB SN4000, which is being offered at a new introductory price as a three-node StarterSAN for $21,000. ICS' first StarterSANs were a 3 TB model for under $12,000 and a 6 TB (two 2 TB nodes) for $15,000. Customers can mix node sizes in a single cluster going forward. Scale plans a 36 TB system within the next six months, according to its vice president of marketing Peter Fuller.

Scale Computing sells its products exclusively through the channel, but Fuller said an ICS design called the iSN1000, based on IBM's System x servers will be distributed through IBM VAR Avnet over the next month or so. That system, according to Scale Computing's website, will be able to scale to 2.2 PB in a single system, closer to the existing enterprise-scale offerings IBM already offers around GPFS.

Taneja Group consulting analyst Arun Taneja said he's intrigued by the connection between IBM and the startup. "IBM is trying to make a huge thing out of GPFS," he said, referring to announcements IBM made of vertically-integrated stacks of storage, servers and software around GPFS last month. "It's consistent with the IBM strategy, which is to make GPFS pervasive – this may be the first example of that."

Taneja pointed out that scale-out systems based on commodity hardware are popping frequently, such as Nexenta's development of Sun Microsystems' ZFS, and others based on software that customers can add to the hardware of their choice. He said Scale will have to do more to distinguish itself in what is becoming a crowded space.

"If the only difference between your product and someone else's product is price, it's only a question of time before your product becomes history," he said.

But Taneja said Scale's expandability is a differentiator, especially the ability to combine nodes of different sizes into a single cluster while using all the capacity of each node. He said small companies also generally look for prepackaged appliances rather than do-it-yourself software-only offerings.

Dig Deeper on Parallel file system

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.