Keeping your business-critical applications running smoothly is a matter of monitoring their behavior, resources, and connectivity. We test and compare top application performance management tools that can help.
Monitoring and tracking the health of today’s distributed applications has become a complex task. While many production applications still adhere to something like the 3-tier model (which separates the presentation, logic, and data tiers) , the actual implementation of those tiers is anything but standard—and can now span on-premises, cloud, and multi-geographical deployments, even within the same application. So while your front-end may be hosted on servers running Amazon Web Services (AWS) , your compute logic might be hosted in Rackspace, and some or perhaps just the most sensitive data files may still be stored in your on-premises data center where you control the firewall.
Even with these complications, however, commercial applications aren’t typically a management problem if the application is popular enough to warrant a dedicated agent from your application performance management (APM) vendor. The real challenge comes from custom applications developed in-house which, more often than not, don’t have the same hooks as the commercial ones. But if your business heavily relies on such a custom application, you’ll need to pay close attention to this roundup of APM tools.
One of the primary functions of any APM tool should be to identify any network event or infrastructure condition that is having an adverse effect on performance. This could come from a multitude of sources and can frequently be traced to third-party products not under the control of the application owner. Most commercial applications—databases, email servers, and Web servers—have some type of monitoring interface for both direct and programmatic use. Any APM tool should have the ability to connect with these commercial applications and monitor their status.
Understanding the design of an in-house application with enough depth to include all external connection points is a critical factor when deciding on the right performance parameters to monitor. For some applications, it should be possible to simply monitor the connection points that would exist to a database, a Web server, or a Web service. Deeper application inspection typically requires recompiling and linking the code to a DLL provided by the monitoring tool. This may not be a viable option for some custom applications but typically provides much better insight into the inner workings of an application.
Quantifying the level of monitoring required for your mission-critical applications is a great place to start an evaluation. If your organization only has a few websites plus a database back end, you probably won’t need some of the high-end features and shouldn’t have to pay for them. Some of the products in this roundup are Microsoft-specific, so you’ll need to look elsewhere if you use a database other than Microsoft SQL or an open-source component. If you do have a complex, multi-tier application and consider it mission-critical, then you’ll want to take a long look at some of the tools with the deeper application-monitoring features.
Key features for managing and monitoring applications start with the configuration process. Connecting to well-known commercial applications should not take more than a few clicks plus typing in administrator credentials. Support for the most popular commercial databases including IBM DB2, Microsoft SQL, and Oracle should be a given as should coverage for the more popular open-source databases including PostgreSQL and MySQL. Increasingly, you’ll want to expand that list even further to include the rapidly growing landscape of cloud database offerings such as Amazon SimpleDB and Microsoft Azure SQL Database.
Configuring and enabling alerts triggered by specific application events is one key capability every tool should have. The alerts need to be able to trigger the notification of a responsible party by using either email or SMS or even integration with a dedicated alerting tool such as PagerDuty. An added plus here would be the ability to automatically remediate certain problems based on a set of predefined workflow steps.
Digging under the covers of a Web-based application requires the monitoring and decoding of traffic between the different tiers. Having the ability to capture that traffic and then decode it to track down problems is a feature typically found only in the higher-end tools. Presenting performance graphics for individual SQL queries or database updates brings an entirely different dimension to basic monitoring. At least one of the products tested (Stackify) makes this possible at a very reasonable cost.
Log file analysis has seen a huge uptick in interest in recent years. Knowing what entries to pull from an application log and then turning that data into usable information typically requires a log analysis product. Stackify offers an add-on to their basic management tool to do just that. Other products in this roundup provide some log-file analysis capabilities at differing levels built in to their basic product, while others offer that functionality as a plug-in.
It should be stated at the outset that the prime objective of this review was not an in-depth look at the finer nuances of monitoring application performance, but more of a high-level look at the tools available to help track applications critical to the organization—and the network resources and conditions critical to them. The starting point of this evaluation concerns the level of difficulty required to set up and configure the monitoring of both Commercial-Off-The-Shelf (COTS) and custom, in-house-developed applications.
Adding the monitoring capability to a custom application should be as minimally intrusive as possible and well documented to include sample code. Some of the products reviewed only require you provide a few lines of code while others need more extensive modification. If your application already has some level of error logging, you may not need to make any changes at all.
Reporting of applications can fall under the category of meeting service-level agreements (SLAs) for specific applications. Building custom reports with specific queries need not be difficult. Exporting those reports to a variety of file formats should also be an easy thing to do. Dashboards should have the same level of flexibility and customization. Graphics presented depicting errors must allow an administrator to determine the problem without a lot of digging.
