Startup Refreshes Data Caching

Silver Peak Systems offers a new technique for caching data that is more comprehensive and less prone to delivering stale data to remote users than existing technologies.

A startup hopes to challenge the old guard in the wide-area file services arena with a new technique for caching data that is more comprehensive and less prone to delivering stale data to remote users than existing technologies.

Silver Peak Systems Inc. last week launched its new Network Memory technology, which stores data at the byte level and employs unique pattern-matching algorithms to recognize previously requested data. It localizes information and transmits only changed data between locations while maintaining centralized control of servers and storage.

Unlike traditional file caching, Network Memory does not rely on file names to recognize previously requested data or require any reconfiguration of clients and works transparently across any application, including video streaming and voice over IP.

Like WAFS, Network Memory employs appliances, which Silver Peak calls NX Series devices, at remote branch offices and a corresponding appliance at the data center. And like WAFS, it is aimed at boosting the performance of servers located in a central data center that are used by remote workers across a WAN, allowing IT to consolidate servers and storage to reduce operational costs.

"Imagine if routers processing packets flowing between headquarters and a branch looked at and remembered all the data that flowed through them before. And then when they saw the same patterns, they could refer to it, knowing there is a corresponding device thats seen and knows that data," said CEO Rick Tinsley in Mountain View, Calif.

Although Tinsley said he believes Silver Peak will compete with WAN optimization vendors such as the former Peribit, now a part of Juniper Networks Inc., and Riverbed Technologies, at least one early user said he believes the companys LIN (Local Instance Networking) approach is more comprehensive, according to Patrick Wilson, chief technology officer and co-founder of IT Surgeons LLC, in San Ramon, Calif.

/zimages/5/28571.gifClick here to read more about Junipers acquisition of Peribit.

"WAN optimization fits a particular bill, but its no longer the whole story. Silver Peak manages to get the performance benefits of WAN optimization, file-sharing optimization and application optimization in one box," said the former Peribit user.

When a request is made for data that has already traversed the WAN links and been stored locally, the central appliance sends a reference of that data back to the local appliance, rather than the actual data, which is then served from the local appliance. Any changes to the data are also transmitted.

"When we did our initial testing, if we had a file that normally took 4 minutes to go from Point A to Point B, when we sent that same file with the appliances in, it only took 2 and a half minutes. When we went to send the same file thats now [capable of being referenced], it went from 2 and a half minutes to 15 seconds," said Eric Beasley, senior network administrator at Baker Hill Corp., in Carmel, Ind.

Silver Peaks NX Series appliances also perform compression, employ techniques to cut down on TCP/IP "chattiness," include quality-of-service features and provide IP Security encryption.

The NX Series appliances are available now. Prices range from $9,995 to $49,995.

With the market for WAN or application optimization maturing quickly, Silver Peak will have to do more than demonstrate it has superior technology, said analyst Joe Skorupa, of Gartner Inc., in Stamford, Conn.

"Todays business environment is changing, and its more about who can solve the problems in large deployments. Broad market acceptance will require them to develop a partnership with at least one medium-to-large, credible vendor," Skorupa said.

/zimages/5/28571.gifCheck out eWEEK.coms for the latest news, views and analysis on servers, switches and networking protocols for the enterprise and small businesses.