| QUESTION & ANSWER |
Garbani: Packaged apps may not matter, but networks do |
 |
By Jim Rendon, SearchNetworking.com News Writer
30 Oct 2003 | SearchNetworking.com |
|
|
Packaged applications have become so entrenched in the enterprise that they no longer provides companies with a competitive advantage, according to Jean-Pierre Garbani, a director with Cambridge, Mass.-based Forrester Research. Instead of deriving advantages from applications, enterprises will find that networks are the key differentiator. Businesses that deliver applications the best will be the ones that gain the most competitive advantage. Here, Garbani details some of the ways that business can get the most from their networks.
At this year's Networking Decisions conference, which runs Nov. 5-7 in Atlanta, Garbani will offer even more insight into how to best manage networks. You say that IT is no longer a differentiator. Why?
Jean-Pierre Garbani: That's from Nicholas Carr's Harvard Business Review article, 'IT doesn't matter.' He said that we are moving toward more and more packaged applications and inexpensive hardware. In that context, you cannot dream of gaining [a] competitive advantage from technology. If you buy software from Siebel [Systems Inc.] for customer care, there is nothing preventing your competitors from doing the same.
Then how can companies use technology to gain a competitive advantage?
Garbani: The advantage is in how well you deliver the service. For instance, assume you have Siebel in Company A and Company B. If the application is available 100% of the time with a three-second response time in Company A, but in Company B it is available 99% of the time with a 10-second response time, then Company A is more competitive. Systems management becomes the most important function.
What tools are available for ensuring delivery and management?
Garbani: There are a certain number of fundamental tools to manage servers or applications. All of these tools are useful, but what is missing is the ability to bring them all together into a view that is representative of the service. To do that, you have to create models, or maps, of the traffic flow. That won't replace what we have today. It will supplement what we do.
There are two kinds of players in this space: the traditional camp, such as Hewlett-Packard [Co.] and Computer Associates [International Inc.], that are working from the bottom up. They say we have the tools to capture the data; now here are the tools to interpret the data.
Other vendors say you already have these data capture tools, and we will provide you an additional tool that helps you interpret the data. The big question is, 'How do you create these models?' Do we have an effective way of mapping applications to infrastructure? Maybe. Tools exist to do it by hand, but it is difficult. You have to have a detailed knowledge of the infrastructure, and that will only work in organizations up to a certain size. What we really need is a tool that does application discovery that will give you the map of how application traffic moves across the network. In the next 12 to 24 months, we may see a solution.
Mapping is part of the process of linking IT to business process. Why is that important?
Garbani: You have two perceptions of service: one is a user sitting at a desk trying to use an application, while the other is the IT operation's perception of devices. But knowing about network devices does not give you a concept of service, just whether routers or switches are up or down or working within specifications. Often you hear that users are unhappy with performance, yet everything appears to be working just fine. You need to reconcile those two views somehow. You need to create a model of the service so you can understand what is involved in delivering that service, what network segments are involved, etc. Then you can begin to see the business impact of a problem on the network.
What promising technologies are being developed in this area?
Garbani: One company, called Relicore Inc., has a product that targets configuration management. It is an agent that discovers the servers integral to a given application. But it doesn't tell you a lot about the network. They still have some work to do.
Computer Associates is trying to put together a product called Sonar that it says is based on protocol analysis. It captures data on the network and determines who is talking to whom for different applications. It can then talk to Tivoli. It can do that for Java 2 Platform, Enterprise Edition (J2EE) applications. The problem is that it may not work well for legacy applications.
Microsoft is talking about an approach that would create a file that records the application's dependencies and follow it throughout its life. It's not a bad idea, but it is limited, since it would apply only to new Microsoft applications.
FOR MORE INFORMATION:
Meet Jean-Pierre Garbani at Networking Decisions.
Learn how Critical apps rely on enhanced sniffers.
|
|
|
|
