Networked applications are nothing new. What is new is the shift in the management focus from availability and event management to application performance management. Network availability/reliability is no longer the pre-eminent problem it once was. The topics of interest now are the service levels experienced with the increasing number of networked applications (VoIP, MPLS, etc.) appearing in the enterprise. In general, short of a physical break from whatever cause, the network is pretty much available and accessible, 24/7, x365.
The problem with all these networked applications, however, is maintaining -- in the face of escalating demand -- acceptable levels of performance, whatever the service level metric (such as the performance of an individual service application or overall access for a class of critical end users). IT is spending more and more of its time analyzing and tracking what is happening across the network, at endpoints and in-between, to keep service performance at acceptable levels. What can be done?
A strategy for application performance management
The solution is to work from a coherent strategy to understand what needs to be done, assess what you have today that can help, and then determine what you need to add or change to accomplish the task. Only rarely is such decision making and planning done from a completely clean slate without existing tools or with a blank check to rip and replace existing tools with a totally new solution set. Therefore, the next best starting point is to plan on using as many of the existing monitoring and management tools as possible.
1. What needs to be monitored?
First, you need to define the scope of what needs to be monitored and managed when dealing with networked applications. To provide satisfactory performance levels, you must first be able to identify an array of problems that can have their source in the applications themselves, in the servers on which the applications run, or in the network. In the network, devices, connections and configuration can all have an impact on the end-user service experience. Next, you must determine how to resolve these problems.
For now, we'll concentrate on assessing monitoring and management needs. At a minimum, you must plan to monitor:
- End-user experience: You need to know whether there are response time problems before end users call the help desk or service desk to complain. You need to know whether a business application or service could fail or slow dramatically, threatening business revenues.
- Servers at both ends and the devices in the network: You need to know whether everything is configured correctly. Are servers performing correctly? Do performance level trends indicate an emerging problem? Is anything near a saturation point?
- Traffic through the network: You need to know which applications are consuming the most resources. Is the routing appropriate? Are some links over- or underutilized?
2. Which monitoring tools do you have now?
Once you understand what you need to monitor, you can move to inventory and assess which tools and application resources you have available to do that task today. Then, determine what data and information is currently collected. This information can provide insight into what is happening in the operational environment and how it will affect business service delivery and performance commitments. Knowing which tools and data you already have available will allow you to determine the additional resources you need in order to perform the overall management task.
Remember, you need to be able to understand not only the performance of the infrastructure but also the interactions and -- this is important -- what the impact is on the service delivery as experienced by the business or business customer. Your requests for additional resources will be more effective if you can report to the business managers that -- unless a router is updated or bandwidth added or the workload reduced -- their biggest customer won't be able to enter orders to meet end-of-quarter goals.
In the next tip in this series, we'll continue our examination of what is to be done to achieve top-level quality of service with networked applications. We may even look at what some vendors have to offer. In the meantime, email your questions, suggestions, requests or comments to email@example.com.
About the author:
Richard Ptak is founder and partner at Ptak, Noel & Associates. He has more than 30 years of experience in systems product management. He was VP at Hurwitz Group and D.H. Brown Associates and worked at Western Electric's Electronic Switch Manufacturing Division and Digital Equipment Corporation. He is frequently quoted in the trade press and is author of Manager's Guide to Distributed Environments.