GCN LAB REVIEW
Steelhead appliance can cut latency out of the cloud
Device works to eliminate unnecessary traffic between an application and its users
- By John Breeden II
- Apr 01, 2011
Many federal agencies have their heads in the clouds. And those that don’t really want to.
It seems every other story these days is about how a federal agency is working on cloud computing, running studies about the feasibility of it or thinking about offering it to other agencies.
There are many advantages to moving to a cloud computing environment. It saves data center space, makes tech support easier, reduces overhead and repetitive programs on the network, and enables a level of collaboration that’s not easy to duplicate through other means.
However, there also are a few disadvantages, which some unprepared agencies might not expect. The biggest one is that, no matter how great your network is, there will be latency problems the further a user gets from an application. In a cloud environment, that problem is multiplied.
The problem is compounded by programs that are designed for local-area networks and send a lot of data back and forth or are generally inefficient in their programming. Although that might be fine when working on a LAN, it will cause problems in a cloud. If latency gets too high, users will really feel it, and that reduces many of cloud’s inherent advantages.
The Steelhead appliance — there is also a virtual software version — from Riverbed Technology uses wide-area network optimization techniques developed to reduce or eliminate latency in the cloud.
Steelheads inside a cloud typically sit between routers and switches, and all network traffic runs through them.
The first thing they do is scan all byte-level traffic crossing the network. A good example is an agency logo. That logo might live in thousands of e-mail signatures, on PowerPoint slides, and in countless memos that cross through or are called up in the cloud every day. Just sending the full logo back and forth increases bandwidth use and, thus, latency. The Steelhead device will create a byte reference for that logo and use the reference each time an application needs it. Multiply that savings by all the typical network traffic, and you end up with big savings and a more efficient cloud.
The Steelhead also works directly with common programs to eliminate unnecessary traffic between an application and users. Most applications were never designed for the cloud, so they do a lot of unnecessary talking. The Steelhead engineers work with programs to streamline their use when running traffic through the appliance.
Joe Ghory, product marking manager at Riverbed, said that when optimized with a Steelhead, users of Microsoft SharePoint will find that pages load 18 times faster than when trying to run it natively in a cloud. This optimization works with often-used business programs, including file sharing, Exchange, Lotus Notes, Web, database and disaster recovery applications.
Adding a Steelhead to a network is also relatively inexpensive considering that the time and bandwidth saved is unlimited and grows as more people and programs interact in the cloud. It costs about $5,000 to use Steelhead to optimize a small office with a few users working in a cloud. For a data center with thousands of users, the cost is about $235,000. But that is a one-time fee, and the appliance can receive updates as new programs are optimized, so it’s quite a good deal for long-term cloud optimization.
Any agency thinking about moving to a cloud environment or is already there and experiencing too much latency should consider implementing Steelhead appliances to get things under control. Only a fully optimized cloud can offer the advantages of this new type of computing architecture.
Riverbed Technology, www.riverbed.com
John Breeden II is a freelance technology writer for GCN.