This is part one of a two-part series. Part one tells about monitoring and management. The second part will explain what tools are available to use to achieve it.
End-to-end monitoring and management is used to gather insights into all aspects of an application instance including the OS, network, middleware, database and the application itself.
End-to-end management is successful when system, network and application management are combined in an effective way. An important part of a systematic approach is coordination by IT personnel using procedures and data. Software tools without skilled personnel and procedures to follow are simply not that useful.
Management disciplines based on system, network and application management are mature but, despite their maturity, there are challenges. Many organizations have realized that the IT personnel supporting each of the three main areas must work together to provide a level of human integration. This is a challenge that some organizations have yet to fully address, as it involves a kind of organizational human engineering that requires skilled personnel that are not readily available.
Another challenge is the monitoring and management of software. Different tools can have different displays, message formats and conventions. Without integration at the software level, effectiveness can be hampered. It is possible to have success without one integrated toolset, but it’s not ideal. This shortcoming can be addressed with software that provides at least a rudimentary level of integration, however implementing this software engages resources and requires careful testing.
End-to-end monitoring and management helps to make IT more effective. Without it, you are constrained when problems occur. Insights at just the system and network level are often not enough to solve application problems. Today’s middleware and database technology plays such an important role that integrating tools that monitor and manage these components is essential. History has shown that bring together these elements in a meaningful way is a tough challenge.
How to Implement
Chances are you already have a monitoring and management approach that is being used. If you do and problems are still difficult to solve, then perhaps a tune up is required. If you don’t have a monitoring and management regime, then it is probably time to rationalize an approach, write it down and socialize your main ideas. Here is a framework that should help you think through the challenges:
- Use a balanced approach to monitoring. Monitoring is necessary. It can be proactive by using software to check on the key state of important system, network and application resources, or it can be passive by scraping logs for important messages, but it should be done. If you fear agent-based technologies, then deploy agentless tools. If you don’t have confidence in agentless technologies, then employ human monitors to run the important paths through the application logic. I worked for a company that used this approach when there was no other cost-effective way to ensure a service-level agreement for a critical application. There are so many ways to handle monitoring, no matter your budget or concerns, you should find a way to implement a balanced monitoring approach.
- Use a management toolset that provides the necessary level of help for your support personnel. Management tools are an important part of the end-to-end monitoring solution. The tool should be aligned with the user. It is an oversimplification but it isn’t the best choice to give a toolset to level-one support personnel that is a better fit for system programmers. Since it is cost prohibitive to staff the support desk with a system programmer, you should leverage a toolset that makes support staff as productive as possible by leveraging powerful yet approachable tools.
- Use a toolset that fosters the necessary degree of software integration. If you are looking for a toolset then use one that is already integrated, like the Tivoli* products that use a common framework. If you have an existing toolset, put some focus on integration software that will automatically tie together your current monitoring and management software.
- Use approaches that reuse assets like stress-test scripts that can also be used for monitoring. It is easy to get into a rut where you have software that monitors and software that manages, and is perhaps difficult to think out of the boxes that they are in. However, there are some alternatives that are not so obvious, such as the multiple use of test scripts. Many IT departments stress test their new applications using testing software. They create scripts by manually capturing the main paths through the application logic. They then use the software to run many transactions using the scripts in an automated manner until the application processing stalls out. By reaching this break point, developers can isolate the weak part of the logic that is creating the bottleneck. This is certainly useful. To get more use from the same scripts, some IT departments are leveraging their testing investment as a monitoring tool.
- Use approaches that expand typical IT measurements like measuring the end-user experience. Just as some IT departments have not made a commitment to monitoring, others have not attempted to measure the end-user experience. It is one thing to monitor an OS or a middleware component but something else entirely to measure the actual response in the transactions that the end user uses to run the application in production work.
Joseph Gulla is the IT leader of Alazar Press, a publisher of children’s literature.