PGCon2018 - 2.5

PGCon 2018
The PostgreSQL Conference

Speakers
Avinash Vallarapu
Schedule
Day Talks - Day 1: Thursday - 2018-05-31
Room DMS 1120
Start time 15:15
Duration 00:45
Info
ID 1240
Event type Lecture
Track Case Studies
Language used for presentation English

Migration of a 5 Node Oracle RAC to PostgreSQL

Migration of a 5 Node Oracle RAC to PostgreSQL

In this proposal, I would like to talk about the methods and tools implemented to Migrate from a 5-Node Oracle RAC to PostgreSQL.

It was an Oracle RAC with 5 Nodes designed to automatically load balance and scatter the transactions across multiple nodes for a complex business logic. There was a DR(Disaster Recovery Environment) which was a 5 Node RAC again with the ability to take a Read Traffic for ETL and Reporting. There are several built-in features with Enterprise Version of Oracle which are not claimed to be available for PostgreSQL and thus stopping majority of the users to migrate from Oracle RAC to PostgreSQL. I have overcome all such challenges and implemented a Design without using any external tools or Multi-Master solutions in the market.

Target audience:

All the DBA’s and Professionals/Management looking to Migrate their Oracle RAC to PostgreSQL but worried about Performance.

This talk includes :

Here is what I am going to talk about, in brief:

  • Design that replaced a 5 Node Oracle RAC setup with PostgreSQL.
  • Steps to be considered before Migration and Rightsizing your PostgreSQL. – Migration of Business Logic from Oracle to PostgreSQL.
  • Migration of Data from Oracle to PostgreSQL and continuous replication to avoid a Major Downtime. – Load Balancers and Connection Poolers implemented to distribute transaction load in PostgreSQL.
  • Use of Foreign Data Wrappers in this Architecture.
  • Tools we used to enable Automatic Failover to make the Application connections redirection seamless.
  • Issues faced during Migration/Upgrade. – Implementation of Plprofiler to understand the performance bottlenecks.
  • Snapshotting Solution we implemented to build the same feature as Oracle History Snapshotting to provide reports equivalent to AWR in Oracle.