Performance. Problem. Two words DBAs never want to see together. Yet avoiding (or eliminating) database performance problems can eat up hours, days or weeks of an administrator’s time. We have a ton of resources available on ToadWorld. And in previous a blog post we covered the performance objectives that are worth watching: accelerated problem resolution, problem prevention and reducing risk during change. Diving deeper, once objectives are outlined it’s important for database admins to monitor performance. It’s the proactive monitoring that really gives a DBA a crystal ball by which to predict when problems might arise. Of the three levels of performance monitoring – application performance monitoring, instance monitoring and transaction workload monitoring – it’s the last that’s the most complex. Transaction workload monitoring is a challenge because of the highly dynamic characteristics of workloads. There are specialized technologies available to streamline and simplify transaction workload analysis, but it’s important that the solution can account for the following five factors: 1) Context : Since database transactions interact with various complex subsystems, such as OS configuration, analysis must measure performance of all the data. Analyzing the aggregated data from the traction and subsystems provides context on “why” performance is degrading not just that’s it is happen. 2) Problem Prevention: When a problem hits an end users, it’s too late. Tracking and comparing historic data with real-time performance helps DBAs stay ahead of problems, critical to proactive performance management. 3) Data Granularity: While it may seem aggressive, a sub-second collection rate for database transactions is actually ideal since transaction context can change several times a second. Collection frequency dictates data granularity and as a result quality of analysis. 4) Measured Overhead: DBAs should look for transaction workload monitoring solutions that optimize data collection – proving the highest quality collection at the lowest possible cost. 5) Scalability and Integrity: Transaction workflow analysis, even considering the factors above, may be smooth sailing until the volume of transactions peaks. Ensure data collection doesn’t slow when additional databases (resulting in more transaction) are added. Download the Dell Software whitepaper, “Proactive Performance Management for Enterprise Databases,” here for more tips and tricks to ensure your databases are “all systems go!”
↧