Optimizing MySQL without SQL or touching my.cnf
At Dropbox, with 1000s of machines running MySQL, performance is important. Among other things, out team is responsible for our MySQL servers being tuned well. In this talk, we will talk about the performance impact of “environmental” MySQL tuning, where the tuning is neither at the SQL level (like creating indexes or using different table structures), nor with setting MySQL parameters (tuning parameters like innodb_buffer_pool_size or innodb_log_file_size even more obscure ones like innodb_lru_scan_depth). Instead, in the talk we will cover the following areas (we won’t limit the talk to these): - Operating system level tuning opportunities - Compiling MySQL: does it worth to build your own from the performance perspective? - Building MySQL with different compilers - Building MySQL with profile-guided optimization - Impact of using different memory allocators - CPU and memory affinity with running multiple instances The experiments we conducted here helped us to make more educated decisions about how to run MySQL at Dropbox.
Maxim is a Site Reliability Engineer at Dropbox since December 2015. Being a part of database team, his main focus is on automation platform making DBAs life easier. Before joining Dropbox he was Software Engineer at Yandex.
Storage SRE - databases, Dropbox
Peter joined Dropbox's database team in February 2016, has a strong interest in automation and performance tuning. Before joining Dropbox, among many other things, he was Principal Architect at Percona, worked at Sun Microsystems, specialized there in performance tuning and was a DBA at Hungary's largest social networking site. He also taught many Oracle University MySQL courses. He has been using and working with open source software from early 2000s. He currently lives in Dublin, Ireland with his wife and son.