On 15/12/17 13:57, Aleksander Alekseev wrote:
Hi Stephen,
HA/fail-over is a very broad topic, with a lot of pieces that need to be
done such that I'm not sure it's really viable, but perhaps a precursor
project (synchronous logical replication seems like a prereq, no?) would
make more sense. Or, perhaps, a different piece of the HA question, but
solving the whole thing in a summer strikes me as unlikely to be
reasonable.
[...]
I agree with most of the comments down-thread that this projects
seems to undertake much more work than what a GSoC project could be. I'd
like to chime in with a few other comments regarding other aspects the
proposal:
- I believe we're mixing what HA and replication is. HA is about
promoting a secondary node when the primary fails, preventing two
primaries at the same time, ensuring one master is always up and so
forth. Replication is just a data replication mechanism, and that
*helps* with HA, but is not HA nor a necessary pre-requisite (actually
you could build an HA solution based on shared-disk). Indeed, HA can be
built on top of SR, logical replication (with many caveats as have been
already commented), disk replication techniques outside of PG and
probably others. Topic is way broader.
- If the proejct's goal is to improve on some of the logical
replication's shortcomings that would be required to make it a useful
replication technique on which HA solutions could be built on top of, my
+1 for that.
- If the project goal would be to augment an existing HA solution to
abstract away data replication techniques (so that SR is not the only
option, but also logical or disk-based replication or whatever) from the
core algorithm, I believe this is also a very sensible GSoC project.
- To add a proxy layer to any existing HA project is IMO a project on
its own. I don't even see this as a GSoC feasible project.
Cheers,
Álvaro
--
Alvaro Hernandez
-----------
OnGres
