The hardest part of some projects -- especially those involving multiple data sources, formats and interdependencies -- is the finish.
Data-center-wide integrations require keen knowledge of all the IT infrastructure monitoring tools involved, including data formats, API use and so on. The setup must be carefully tested even if the IT team and/or consulting company followed best practices. And the integration project is never completely done, as interconnections are easily broken when patches, updates and new systems are introduced.
Test behaviors between tools
Regardless of the integration mechanism -- API, plug-in, agent, email, simple network management protocol or even out-of-the-box compatibility -- test each element thoroughly to ensure that a tool accesses data as expected. This is crucial when the connection method depends on specific configuration details like a server email address, IP address or protocol.
Integrating two or more IT infrastructure monitoring tools can be tricky, and tests show that you got it right. Test behaviors between the tools to verify they behave as expected. If you integrate Nagios monitoring and management alerts into the Spiceworks help desk tool, for example, Spiceworks remains the principal ticketing system in the enterprise. Configuring the email and parser code in Spiceworks is difficult depending on common message elements from the sending system. Once the setup is finished, trigger an email from Nagios to test if Spiceworks acknowledges and acts on the alert. If not, re-examine your integration work.
Don't just check that it reads data successfully; make sure the tool correctly processes and reports on data. If not, a normal log data entry could trigger false positive alerts, or performance data could skew graphical reports.
There is such a thing as too much communication. Agents, coding with software development kits and other integration mechanisms don't always guarantee good performance. Busy networks and high volumes of data or message transfers cause problems. Measure the volume of data or messages and the various sources to check if a single tool can handle everything properly. This takes a bit of testing and refinement in complex environments.
Even the most flexible and highly extensible tools do not guarantee interoperability with other tools or systems. There are so many potential combinations of hardware and software that you're bound to encounter a tool that won't work properly with another. Try working with the tool vendor to create a viable plug-in, script or other integration element. If you can't, redefine or shelve that component of the integration project.
Don't let change be disruptive
Each connection and integration method creates interdependencies that are easily disrupted by patches and upgrades.
Consider the Nagios-to-Spiceworks integration where Spiceworks parses common text in the subject line of Nagios email messages. If an update or setup change alters how Nagios generates subject lines, then Spiceworks might fail to parse the Nagios messages. While Nagios changed, Spiceworks is the tool that requires a fix for the integration to work.
Upgrading to new versions can also cause problems when tools receive and parse data from each other. If one tool's latest version moves its log file to a different directory or formats logs differently, another tool that reads the file will need an update to accommodate it.
Server-side changes also strike up problems in a tied-together data center monitoring scheme. For example, Nagios has a plug-in for integrated lights-out (iLO) agentless management on HP ProLiant servers. If HP changes the firmware or system hardware design, the iLO plug-in might need to be updated. Newer servers might not work with older plug-ins, creating issues during a hardware refresh.
Lab testing will ensure compatibility, reveal problems and help with fixes or workarounds. Documentation, change management and testing must accompany any change in the production data center. If changes will break the existing IT monitoring tool integration, reach out to the tool vendor with details and ask if they can assist with troubleshooting and a resolution.
About the author:
Stephen J. Bigelow is a senior technology editor at TechTarget, covering data center and virtualization technologies. He has acquired many CompTIA certifications in his more than two decades writing about the IT industry.
Is it time to consider application performance monitoring tools?
Four steps to finding the right server monitoring tool
Free Windows Server monitoring tools