Friday, March 15, 2019

Monitoring Database in AWS Aurora After Migrating from Oracle to PostgreSQL

Suppose you have an Oracle database on-premise, which you have now moved over to AWS Cloud in AWS Aurora PostgreSQL. 
For your Oracle database, you have been using v$ views to monitor your runtime performance of instance, long running operations, top SQLs from ASH, blocking etc. How do you continue doing that when you migrate your database to cloud especially in AWS Aurora based PostgreSQL?

Well, PostgreSQL provides statistics collection views, which is a subsystem that collects runtime dynamic information about certain server activities such as statistical performance information. For example, you can use  pg_stat_activity view to check for long running queries.

There are various other statistics views too in PostgreSQL such as pg_stat_all_tables to see size of table, access method like FTS or index scan, and so on. There are other views to check IO on tables and indexes and plethora of others.

In addition to these statistics views, Aurora PostgreSQL provides a nifty tool called as Performance Insights. Performance insights monitors Amazon RDS or Aurora databases (both MySQL and PostgreSQL) and captures workloads so that you can analyze and troubleshoot database performance. Performance insights visualizes the database load and provides very useful filtering using various attributes such as: waits, SQL statements, hosts, or users.

As part of operational excellence, its imperative after a database migration that performance is monitored, documented and continuously improved. Performance Insights and the statistics views are great for proactive and reactive database tuning in AWS RDS and AWS Aurora.

No comments: