On Tue, Jan 10, 2012 at 8:56 AM, Ron Somaraju <RSomaraju@masergy.com> wrote:
> David, thanks for the response. I appreciate it. I hope the community understands the benefit of providing the
functionalityto be able to dynamically increase WAL size without having to re-init the cluster (which I think is the
workaround). In this information age an OLTP or OLAP database with an average size of 1to 3Tb size is not uncommon.
Withthat in view an active database could generate hundreds of 16Mb files within a very short time. In my case we are
movinga large active database from Oracle to Postgresql. In Postgres I am seeing an average of 15 to 23 WAL 16Mb files
everyminute.
While I'd love to see a test of how things work with 64M WAL files,
I'd worry about how well tested that particular code path was to use
in production. OTOH small incremental changes are often the best to
test with.