thePlatform is a leading, fast-growing video management and publishing company for broadband, mobile and TV. Since its debut, thePlatform quickly grew revenue, as well as momentum, in the media sector, and has successfully gained well-known customers, including Comcast, NBC-Universal, Associated Press, BBC Global News, HIT Entertainment and CNBC.
The Business Challenge
thePlatform had been using proprietary databases to store meta data for media files, such as locations of the files and the customer information, in the Media Publishing System. Since business at thePlatform has grown extremely fast, the demand for the application has increased dramatically with more customers and more end users per customer. To further scale the Media Publishing System and reduce costs, thePlatform decided to deploy a scale-out database architecture and to migrate the application to MySQL.
The MySQL Solution
MySQL Query Analyzer
It has been a challenge for thePlatform to identify performance problems efficiently with MySQL when one customer experienced longer response times than others for the same service. While the DBA team can "playback" the application load using the MySQL General Log or Slow Query Log, it was almost impossible to spot the problem queries quickly in the log file. Also, the problem query has to run long enough to be picked up by the Slow Query Log.
The MySQL Query Analyzer, with a unified panel that lists all the queries by its execution count, run time and result sets, is an ideal solution to this problem. It captures queries with a shorter execution time, provides complete information for each query, and offers various dimensions to examine the queries. With the MySQL Query Analyzer, thePlatform now has a systematic approach to gather and test the application load, and to improve service response time by finding the worst-performing queries.
thePlatform describes the MySQL Query Analyzer as "a great tool for trouble shooting" and has successfully improved performance by orders of magnitude. In one case where thePlatform was testing the new version of a custom service API, the DBA team noticed that the response times kept increasing during the performance tests but had difficulties identifying the associated query. After obtaining the MySQL Query Analyzer, thePlatform re-ran the same test and sorted the queries by the maximum run time, and quickly found the queries that were causing performance issues.
Once the problematic query was identified, thePlatform immediately recognized a missing index. By simply adding back this missing index, the query time was decreased from 70 seconds to 3 seconds, effectively a 23x performance improvement. Since this API call is executed very frequently, other services also benefited from this fix with reduced response time. This huge performance gain, which would have required several days of effort, only took thePlatform a few minutes to achieve with the MySQL Query Analyzer.