Home > Diagnosing Database Performance Problems When You Aren’t a Database Administrator

Diagnosing Database Performance Problems When You Aren’t a Database Administrator

Deep specialization of IT administrators is a luxury only the largest organizations can typically afford.  Smaller organizations rely on IT administrators with a more generalist skill set because they are—by necessity—responsible for a wide array of different technologies, and there simply isn’t time to specialize in the intricacies for any one of them. Yet modern IT is intricate. Hybrid and multi-cloud deployments, distributed applications, and unwieldy collections of data are no longer the sole province of large enterprises with teams of specialists. They’re increasingly the norm for smaller organizations as well. Generalists must manage storage, networking, and compute across disparate infrastructures; troubleshoot app performance; plan for disaster recovery; and more. There are always intense demands on a generalist’s time, so anything capable of simplifying and streamlining their work has an outsized positive impact. Management, monitoring, and analysis tools can be (and frequently are) the difference between success and failure for generalist-heavy IT teams. In particular, the use of machine learning has made monitoring software a powerful ally of administrators. Let’s look at database monitoring as an example.

Defining Normal

One of the goals of monitoring is to understand what’s normal for an organization’s IT. What does it look like when it’s functioning smoothly? What’s an acceptable fluctuation in performance vs. a cause for concern? These are the types of questions machine learning is well equipped to answer. Database performance monitoring tools collect granular, second-by-second information about resource utilization and wait times to build a picture of normal operations that only becomes more accurate over time. If a parameter strays outside the expected range, IT staff can be immediately alerted and presented with relevant information in an easy-to-understand visual format. But until such an alert appears, they can focus on other tasks without worrying about missing something crucial.

Analysis and Mitigation

Once a problem has been detected, the IT generalist needs to figure out what’s causing it and what to do about it. Generalists are often skilled researchers, but the time necessary to resolve a problem can be greatly reduced through access to database analytics augmented by machine learning. Database performance monitoring tools can detail the problem for the administrator, allowing them to reduce the scope of their research efforts. Database performance monitoring tools allow IT teams to identify the root causes of database issues and presents suggestions for action. Is a particular application struggling with performance? Identifying the problem, however, doesn’t mean much if you don’t know what to do about it.  A useful database performance monitoring tool will need to provide helpful information designed to help walk an IT professional through the process of splitting databases to give the problem app its own dedicated resources. Periodically databases require maintenance (for example, if indexes have become fragmented). A proficient database performance monitoring tool can highlight this and suggest how best to proceed. Is a specific SQL statement causing problems? The database performance monitoring tool needs to be able to advise IT teams on how to tune their tables to highlight efficiencies (such as how many queries would be affected by adjusting a single index) so IT staff can take informed action. To accomplish this, the database performance monitoring tool should be aware of the specific characteristics of all the major enterprise database engines. Each responds differently to particular resource constraints and can be significantly affected by table design, configuration, and other optimizations. It also matters whether a database is running on a physical server or whether it’s virtualized, containerized, or operating in the public cloud. Resource constraints can almost always be solved by tuning, but occasionally the underlying infrastructure will need to be upgraded or expanded. If this is the case, a well-built database performance monitoring tool should be able to create reports showing clear connections between available resources, database performance, and what the end user is experiencing. IT staff can easily share these with financial decision makers, explain the situation, and advocate for necessary infrastructure spending.

Lighten the Load

Powerful monitoring and analytics tools function like an extra person on the IT team, helping generalist IT practitioners keep their workload manageable. These tools can help ease the burden of watching for anomalies, doing root cause analysis, researching solutions, prioritizing actions, and providing reports to share with other business departments. They free up IT generalists to spend time on other vital parts of IT operations. SolarWinds® Database Performance Analyzer provides in-depth compute metrics—CPU, memory, disk, network, and so forth—as well as database-specific metrics. Experience a database performance management solution capable of providing intelligent recommendations based on best practices to help solve and prevent performance issues in hybrid IT environments. Download a free trial of Database Performance Analyzer today.
Joey D'Antoni
Joey D'Antoni is a principal consultant at Denny Cherry and Associates Consulting. He is recognized as a VMware vExpert and a Microsoft Data Platform MVP…
Read more