Getting the most from big pipes

I’ve written a lot lately about integrated devices and the trend to bundle multiple traffic management, security and routing capabilities in a single appliance to help combat branch office clutter.

But not all WAN optimization vendors are moving in the branch-office-in-a-box direction. A company with a different objective is Silver Peak Systems, which continues to sharpen its focus on high capacity WAN links and their distinct performance, scalability and configuration challenges.

Silver Peak this week is due to unveil Version 2.0 of the software that runs in its NX family of appliances and Global Management Systems devices. The new version is aimed at improving application performance over big pipes and simplifying the management of distributed NX devices.

High capacity WAN links -- those 45Mbps and higher -- have characteristics that separate them from smaller branch office connections, says Jeff Aaron, director of product marketing at Silver Peak. They typically involve larger volumes of data, more types of applications, larger numbers of simultaneous user sessions, and more complex network configurations requiring asymmetric routing and advanced QoS policies, for example.

To continue reading, register here to become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

I’ve written a lot lately about integrated devices and the trend to bundle multiple traffic management, security and routing capabilities in a single appliance to help combat branch office clutter.

But not all WAN optimization vendors are moving in the branch-office-in-a-box direction. A company with a different objective is Silver Peak Systems, which continues to sharpen its focus on high capacity WAN links and their distinct performance, scalability and configuration challenges.

Silver Peak this week is due to unveil Version 2.0 of the software that runs in its NX family of appliances and Global Management Systems devices. The new version is aimed at improving application performance over big pipes and simplifying the management of distributed NX devices.

High capacity WAN links -- those 45Mbps and higher -- have characteristics that separate them from smaller branch office connections, says Jeff Aaron, director of product marketing at Silver Peak. They typically involve larger volumes of data, more types of applications, larger numbers of simultaneous user sessions, and more complex network configurations requiring asymmetric routing and advanced QoS policies, for example.

“Branch offices traditionally are going to have basic file, e-mail, and other office productivity tools that are usually client-server applications,” Aaron says. “With bigger pipes, you’re now introducing more server-to-server applications, like backup and replication and SQL. These fat pipes have very different characteristics than their skinny brethren.”

In Version 2.0, Silver Peak enhanced its Network Memory technology for doing disk-based data reduction at high speeds. Data reduction can reduce over 95% of WAN traffic by eliminating the transfer of duplicate data, Silver Peak says.

“Data reduction on these big pipes is pretty difficult to do because you’re required to do lots of read and writes, at very high speeds, and you’re writing to disks, which can take time if you don’t doing it correctly. It can potentially introduce a new latency element that you obviously want to avoid,” Aaron says.

The new version of Network Memory features better pattern-recognition capabilities -- which are particularly useful for improving the performance of applications that do “interactive saves,” such as Microsoft Excel, Aaron says.

“Whenever you modify data in an Excel file and resave it, that change is actually randomly distributed somewhere in the file, in terms of bytes. It’s not in a very predictable space -- it’s not at the end, it’s not at the beginning, it’s just randomly sent throughout the file. So unless you have very granular data reduction that can actually look at those bytes and quickly figure out which bytes are different, you’re not going to be able to detect that as granularly.”

On the management front, Version 2.0 now features a centralized policy engine that lets enterprises use common templates across numerous appliances to make configuration and management of QoS, routing and optimization policies easier.