Visual Data Discovery Monitoring – Next Generation
"Give a man a fish and you feed him for a day. Teach a man to fish and you feed him for a lifetime.”
If you like the idea of having a dedicated technical expert from DataStax on your team that can work 24x7 with you at no extra cost to monitor and tune your database clusters for optimal performance, please read on.
The Thorny Problem of Database Tuning
If you think of the database monitoring solutions in the market, most of them are pretty good in terms of exposing the metrics, and to some extent, have basic capabilities to sort and show the results. However, many of them stop at this point and put the burden on you to make all the decisions without expert guidance. Oftentimes in the field, I have had numerous instances when troubleshooting a performance problem where I wished that an expert sat right besides me, explained what these numbers meant, and guided me through actually fixing the issue.
If you’re like me, I have good news. With OpsCenter 5.2, we are pleased to introduce the new visual Performance Service: a smart companion that not only displays performance metrics, but also intelligently provides guidance on possible ways to rectify found performance issues and reduces the troubleshooting time by highlighting the probable causes upfront.
What is DSE’s Performance Service?
Before we dig deeper into OpsCenter’s capabilities, it is important to understand the roots of the DataStax Enterprise (DSE) Performance Service. In DSE 4.5, we announced the availability of Performance Service for DSE, which was designed to give you better insight into tuning and troubleshooting performance issues in your cluster. This service allows you to automatically collect and organize performance diagnostic information into a set of data dictionary tables.
Using this service, you can get information on various events such as slow performing queries, view latency metrics, collect read/write activities per client, and so forth. While the Performance Service offers great value, up until this point it has been command-line driven in terms of access. Moreover, the exposed stats are left to your interpretation where making decisions is concerned.
How does OpsCenter Work with the Performance Service?
OpsCenter 5.2 (Fig 1) provides brand new visual monitoring of diagnostics collected through the DSE Performance Service. In addition to exposing key metrics, OpsCenter acts as an “Educator” and provides context-specific recommendations on possible causes and potential ways to fix performance issues. The recommendations greatly eliminate a ton of manual work during troubleshooting exercises.
With OpsCenter Performance Service, you can do the following:
- Visually enable the metrics that you want to track and monitor.
- Track these metrics at different times of the day for exploration/analysis purposes and compare the results.
- Configure pre-packaged relevant alerts that will be triggered during the duration of the metrics tracking.
- Receive suggestions and visual cues on what these issues could mean and some potential remedies to addressing a specific issue.
Fig 1: Workflow of Visual Performance Service
These capabilities provide the following benefits to you when troubleshooting performance issues:
- Much of the manual work involved in performance troubleshooting situations is automated, with the end result being a marked decrease (around a factor of 10X ) in the amount of time spent determining the root cause of performance problems.
- The need for custom scripting and scheduling to detect problem nodes is eliminated.
You can go to the “Performance Services” area ( Fig 2) in OpsCenter and enable metrics that you want to monitor. OpsCenter provides a simple user interface that will help even the novice users set up the relevant metrics for performance monitoring. For the complete list of metrics exposed in OpsCenter for DSE, please refer to the documentation for more details.
Fig 2: Performance Services - Configuration
For the purposes of explaining how this all works together, let’s go through a simple scenario of identifying the slowest queries in the cluster. In this example, we have illustrated the workflow where you enable the “Slow Query log” metric.
As you can see from the output below (Fig 3), the Performance Service provides an easy visual interface that shows various slow performing queries in the cluster. In addition to exposing the query metrics, OpsCenter also allows you to trace these queries so you will have a clear indication on the bottleneck causing each query to be slow. In addition, you can also trace custom queries by setting different consistency levels & coordinators to proactively track and identify issues. Furthermore, OpsCenter also displays some common causes for found problems and potential ways to resolve them as well.
This not only simplifies your troubleshooting exercise, but also greatly eliminates the need to maintain custom scripts that require a lot of manual work.
Fig 3: Performance Services - Slow Queries
At DataStax, we want to provide you with everything needed to quickly optimize and maintain the performance of your database clusters. We are confident that the DSE Performance Service coupled with OpsCenter will decrease your troubleshooting costs, increase your productivity and simplify the ease of monitoring your DSE clusters. If you would like to try it out yourself, download OpsCenter 5.2 for free today. The test was conducted on a 20 node Cluster, 2 DCs with 10 nodes each. The number may vary depending on your cluster size and deployment model.
DataStax has many ways for you to advance in your career and knowledge.
You can take free classes, get certified, or read one of our many white papers.
register for classes
DBA's Guide to NoSQL