《causes-of-downtime-in-mysql》.pdf

Causes of Downtime in Production MySQL Servers A Percona White Paper Baron Schwartz Abstract Everyone wants to prevent database downtime by being proactive, but are measures such as inspecting logs and analyzing SQL really proactive? Although these are worthwhile, they almost always identify problems that already exist. To be truly proactive, one must prevent problems, which requires studying and understanding the reasons for downtime. This paper presents the results of a study of emer- The sample data is biased and might not represent gency downtime issues that Percona was asked to the overall population of MySQL users. There might resolve. It sheds light on what types of problems be a high number of website emergencies compared can occur in production environments. The results to the general population. Web application develop- might be surprising. Conventional wisdom holds ers and startup companies often work in a less dis- that bad SQL is the leading cause of downtime, but ciplined fashion than, for example, banks or phar- the outcome of this research was quite different. maceutical research companies. (Percona’s customer In this paper, downtime is defined as time that the list does include those industries, and the sample in- database server is not providing correct responses cludes emergency downtime issues from those types with acceptable performance. Downtime could be of customers.) Therefore, this study might over- because a server has just been rebooted and is run- represent mistakes made by people building ap- ning on cold caches with increased response