As part of Network World’s upcoming IT Roadmap conference, analyst Jim Metzler of Ashton, Metzler & Associates will take a good hard look at application acceleration.
As part of Network World’s upcoming IT Roadmap conference, analyst Jim Metzler of Ashton, Metzler & Associates (and co-author of our WAN Newsletter) will take a good hard look at application acceleration. Making network applications run faster and better can be a thankless job, since network managers usually only hear about performance when there’s a problem. When’s the last time someone in accounting came to you and said, “Hey, those financial applications are flying today, let me buy you a cup of coffee”?
Yet application acceleration is an important, and often strategic, part of the network. In a preview of his IT Roadmap keynote presentation in Boston on March 20, Metzler recently spoke with Network World Senior Editor Cara Garretson to explain why application acceleration matters and to sort out some of the confusing issues around this market.
Why should a network manager care about application acceleration? Isn’t that for the application guys to figure out?
Well, for starters it helps for the IT shop to understand what applications run on the network. If you’re blind to it, you just don’t know, then that gets into network abuse… apparently you can stream Sirius (satellite radio) these days to your network. That consumes a lot of bandwidth, and if users are having difficulty booking sales because someone thinks Howard Stern is funny, that’s a problem.
If you’re shipping big files during the day, maybe you could do that at night, or do a different amount of VoIP than you thought. You need to get an understanding of what’s running on the network.
Okay, but then how do I decide which applications need acceleration?
Well, people are only trying to get certain applications to run better. There are a lot of techniques you can use, but you can’t make everything run better. So you have to decide, of the applications running on the network, what are the three, four, five, six you’ve got to pay the most attention to? What are the business-critical applications, meaning when they don’t perform well it impacts your [performance] review? And what are the applications that are really delay-sensitive? Voice has to be at the top of the list. But the majority of companies have one, two, maybe three delay-sensitive applications. You need to understand what those key applications are.
How, exactly, does application acceleration work?
Well, let’s start with compression. It’s incredibly simple - take something big and make it smaller. As opposed to sending 10 of something, I can send 20. Then there’s caching. Let’s say some company’s branch sales office is in Minneapolis. Every time they access a large file it has to cross the network - you can see where that could slow things down. But what if they stored some of the information locally on their machine? Then all that stuff doesn’t have to go back and forth. They can cache it locally… could be on your PC or a device or on a proxy server in the office. All of this makes the application run better.
It’s taking stuff off of the network so it won’t interfere with other stuff. The difference is you keep reducing how much is going across the WAN and require less stuff in general to be sent.
Do the applications being accelerated need to be aware?
The applications don’t need to be aware. However, there are some approaches that are very application-specific. For example, let’s say one solution understands how Oracle Financials work. That’s very application-specific. But in that case, you’re going to deploy a technology that won’t do a thing for some other application, but makes this application run very well. Very simply put these [application-specific] solutions understand the messages that have to go back and forth as part of Oracle, and they apply different things just to the Oracle application.
You can apply these same kinds of techniques to the two or three applications you have that are strategic. But the vast majority [of application acceleration products] want to do just one thing with an appliance.
So most of these offerings are appliances that work with all applications?
Well, the market is evolving. Early on it was all about hot boxes. Three years ago that was fine, but we’re beginning to see the emergence of systems where vendors who used to do one thing say, “Don’t think of me as an appliance. I’m a system.” We’re getting to the point where people don’t want five or six or eight appliances littering their branch offices. That’s best-of-breed vs. a system; it’s an issue we have for everything we do. A system does everything kind of okay. But if you work at a company where one problem is compression, but over time when you want to do more, will [that box] do an okay job at that too?
What’s going on with all these application acceleration vendors?
I think we’re at the cusp of Phase Two in this market. We’ve moved from hot boxes to more of a system or platform approach, and part of the challenge is it has got to be able to do a good job on basically everything. I think we’re really at the second phase, then the question is what goes into a platform, and how is it managed?
We’re seeing consolidation and expansion, which is really fun. Juniper bought Peribit and Redline, Fluke bought Visual Networks, Citrix bought NetScaler - vendors buying people, but at the same time you’re getting start-ups; we’re seeing consolidation, but at the same time there’s continued interest and VC funding in new companies.