Best practices for running MySQL on production - Vaibhav Upadhyay
Download as PPTX, PDF0 likes392 views
This document discusses best practices for running MySQL in production environments. It covers the importance of backups, having high availability (HA), and troubleshooting MySQL issues. Specifically, it recommends: 1. Using both logical and physical backups, with logical backups being storage efficient but physical backups being faster to backup and restore. 2. Implementing basic HA to avoid single points of failure, provide fault tolerance, and reduce unplanned downtime. However, HA cannot be used as the only backup strategy. 3. Defining problems clearly, checking resources and historical usage, comparing query behavior, and reaching out to the community for help when troubleshooting MySQL issues.
Best practices for running MySQL on production - Vaibhav Upadhyay
- 2. Best practices for running MySQL on production
- 3. Agenda: Backup tool for MySQL Importance of having basic HA MySQL trouble shooting
- 4. Little about me: Working for Techmojo solutions Leading DBA and Production support engineering MySQL user and try to contribute to forum and gain knowledge I thank MyDBOps and Karthik for inviting me!
- 5. Backup tool for MySQL: # Disadvantages Logical : Comparitively takes longer time Takes long to restore For certain size it's not practical Physical : Storage intensive Nobody can save you, but your backup can! Types of Backup Logical and Physical # Advantages Logical : simple commands for backup and restore storage efficent Physical : Faster to backup Faster to restore Also crons and config files and may be mysql users in some cases # Monitoring backup # Test backups
- 6. Importance of having basic HA: Why HA? HA is your friend. :) 1. Fault tolrant architecture 2. Avoid SPOF (Single point of Failure) 3.Your replica is for protecting against the downtime due to HW, SF, DC, Network failure but not for data loss! In other words replica can't be your backup strategy! 4.It protects from unplanned downtime and helpful to reduce the unplanned downtime 5. High availability = High cost and not having HA can cost more than HA cost !
- 7. MySQL trouble shooting: Define your problem (it may evolve over the period) Have concave approach 1. Start with resources 2. Check the historic resource utilization 3. Check the pattern if any 4. Check for latest changes / deployments 5. Compare the query behavior 6. Hunt for newly added queries (monsters) 7. Define what is your benchmark 8. Know your application well Tools can be helpful PMM Percona-toolkit Reach out to community for help
- 8. Run it like you OWN it!! Keep a watch on access Audit privileges periodically Backup test plan HA health Performace reports Table growth Top 10 alerts Top 10 slow queries Robust SDLC system to filter thing before it goes out to Prod