veyepar: Geographically distributed multi-master replication with PostgreSQL and BDR

Author(s):	Craig Ringer
Location	Medium Lecture Theatre CB11.00.401
Date	jan Thu 25	Days Raw Files
Start	14:35	First Raw Start	14:09
Duration	0:45:0	Offset	0:25:32
End	15:20	Last Raw End	15:39
Chapters	00:00 0:04:25 0:34:24
Total cuts_time	44 min.

http://lca2018.linux.org.au/schedule/presentation/75/
raw-playlist raw-mp4-playlist encoded-files-playlist host archive public tweet mp4 svg png

assets release.pdf Geographically_distributed_multimaster_replication_with_PostgreSQL_and_BDR.json
logs
Admin: episode episode list cut list raw files day marks day marks day image_files

State:

Locked:

clear this to unlock

Locked by:

user/process that locked.

Start:

initially scheduled time from master, adjusted to match reality

Duration:

length in hh:mm:ss

Name:

Video Title (shows in video search results)

Emails:

email(s) of the presenter(s)

Released:

has someone authorised pubication

Normalise:

Channelcopy:

m=mono, 01=copy left to right, 10=right to left, 00=ignore.

Thumbnail:

filename.png

Description:

markdown

Keeping your data in more than one place is vital for business continuity, and databases are no exception. Backups are great, but they won't help you when your normally-connected EU, US and Australian operations all want to continue operating during a major WAN outage. This is where multi-master replication systems come into their own - but they also come with often overlooked costs and downsides. Bi-Directional Replication (BDR) for PostgreSQL can help solve some users' availability challenges in a globally distributed environment... if it's applied correctly.

Multi-master is not magic high availability secret-sauce that can be applied to every problem, no matter what some major vendors' sales teams would like you to believe.

Learn about when to use - and not to use - multi-master replication in general, and BDR in particular. When is it better to stick to a single-master system with failover? If you do select a multi-master system, what kind is best for you? Do you need something partition- and latency-tolerant, or something tightly coupled and transparent? Shared-mostly or shared-nothing? These details will have a major impact on whether a given system architecture can meet your needs, and understanding them will help you set better requirements during a design and product selection process.

Comment:

production notes