Webinar: BDR – Latest Features & Updates [Follow Up]

BDR is a rapidly evolving project — and has been generating great interest from the Postgres community. In order to give further insights into BDR, how it works and where to best apply it, 2ndQuadrant organized a live webinar on 15th January 2020. 

The BDR – Latest Features and Updates webinar was hosted by Simon Riggs (CEO at 2ndQuadrant), with the aim to bring everyone up-to-speed with the latest BDR features, and give a comprehensive analysis of its architecture and cutting edge use-cases. 

This webinar covered the following topics:

• Introduction to BDR (Bi-Directional Replication) and BDR fundamentals
• BDR Architecture and latest features available in BDR
• AlwaysOn – Advanced high availability and rolling upgrades
• Disaster Recovery, Latency and Worldwide Clusters with BDR
• Latest use-cases for BDR
• Making applications compliant with BDR

Those who weren’t able to attend the live webinar can now view the recording here.

Questions that Simon couldn’t respond to during the live webinar, due to time restrictions have been answered below.

Question: Is there overhead in inserts or updates with BDR always on? in percentage, please.

Answer:  There is no foreground overhead for INSERTs or UPDATEs – changes are written to the transaction log and then replicated from there. That is why we say that the replication is very efficient. Applying the changes on other nodes has a cost, but that is generally much less than the original action. Various replication modes add latency, but very little resource overhead.

Question: Can we configure BDR between heterogeneoushyterogeniues clouds like AWS(node1) and AZURE(node2)?

Answer: Yes, we can support Multi-Cloud deployments

Question: How do you distinguish BDR from other scaling technologies, e.g. CitusDB?

Answer: BDR has better availability

Question: Every node/pod can have its own replica with its own streaming replication? A pg_basebackup can raise the amount of wal-files above 100gb?

Answer: Yes, if needed, but we don’t recommend that architecture.

Question: Can we handle table level unique value?

Answer: Yes

Question: Merge Replication Possible?

Answer: "Merge Replication" is a term used by another replication system. It does look like the description is very similar and I’m not aware of any differences, so Yes.

Question: Do you implement backup for nodes on BDR?  Is it necessary to run a backup on only 1 node or do we have to run it on all nodes?

Answer: It’s best to run it on multiple nodes, to more quickly regenerate lost data in each region

To stay updated on upcoming webinars by 2ndQuadrant, you can visit our Webinars page.

For any questions, comments, or feedback, please visit our website or send an email to [email protected]om.