Application impact analysis prior to application deployment is critical for successful network performance. The last several tips in this series focused on developing a strategy around performance management and areas of focus for managing applications. These tips focused on the current application mix and did not address the deployment of new applications on the network. Rightsizing networks for application support is critical. Providing too little bandwidth leads to poor performance and specifying too much is costly. For this reason, application impact analysis before deployment is critical.
Application impact analysis or network application readiness both refer to the process of trying to define what impact an application will have on the network prior to that application's being deployed. Deploying new applications can put a tremendous strain on the existing network in terms of both bandwidth and performance. Understanding this impact and rightsizing or optimizing the current network before deployment can greatly influence the ability to deploy the new application transparently and effectively.
The approach to application impact analysis starts with the business unit or groups that the application will serve. It is critical to understand the business function of the application as well as the classification of the application. Understanding key factors such as whether or not the new application will be distributed enterprise-wide or limited to a certain demographic of users will provide information regarding infrastructure impact.
Once the scope of the application is understood, the next factor to consider is scale. Application flows represent a transaction between a client PC and a server. Each transaction puts a demand on the underlying network. This demand increases with each user who will utilize the application. This brings into play the user demographics at each site and the amount of usage from those users.
At a very simplistic level, the consumption rate or demand of an application is the number of transactions times the amount of bandwidth per transaction times the number of users. Using this simple formula, you can attempt to estimate the amount of bandwidth required at each site. However, there are some missing components. Just exactly how many transactions does an application require -- and how much bandwidth is used for each transaction? When developing the impact analysis, the application is not yet deployed on the network, so there is no way to capture the transactions and bandwidth to measure the impact. In some cases, the application developers or the application vendor can provide some input as to the types of transactions and bandwidth estimates.
The ideal environment for capturing the transaction and bandwidth numbers is in a lab. In almost all cases, applications can be run between a client and a server in a lab environment. Network analyzers can be placed on the network to capture and break down the bandwidth numbers associated with the transactions. This information and the data captured, along with the user demographics, can be used to calculate the bandwidth impacts on your network.
Some vendors will provide guidance on the impact of their applications, but all you get is guidance. True analysis and testing is the only way to understand the real impact. The process of impact analysis should be built into the application development lifecycle in order to ensure that infrastructure readiness can happen in parallel with the application development process.
About the author:
Robbie Harrell (CCIE#3873) is the National Practice Lead for Advanced Infrastructure Solutions for SBC Communications. He has more than 10 years of experience providing strategic, business and technical consulting services. Robbie lives in Atlanta and is a graduate of Clemson University. His background includes positions as a principal architect at International Network Services, Lucent, Frontway and Callisma.