[Postgres-xc-general] Few queries from Varprasad

[Ashutosh B. <ash...@en...>]

Hi All,
One of my colleague in Sybase, came across my
article<http://www.linuxforu.com/2012/01/postgres-xc-database-clustering-solution/>in
Linux For You magazine and sent following queries. I thought it better
to discuss on XC general forum.

Varprasad was earlier working with Sybase and has long experience with HA,
Shared disk cluster and RDBMS in general.

1) what are the HA aspects of various component failures? What is current
HA recommendations for various failures (coo, datanode or gtm etc)?
2) why GTM is desinged as centralized server? Is it possible to make it
work in distributed manner so as to remove the bottleneck and SPOF?
3) what is the expacted scale-up for very large number of nodes 50-100, esp
considering that the scale-up reduces with larger number of nodes.
4) Is GTM used for repeatable reads as well, apart from the Write
transactions?
5) Do all coordinators need to be alive to get this setup working esp for
DDLs. We observed that connection pooling component gives error when the
DDL is fired after a coordinator crashes.

-- 
Best Wishes,
Ashutosh Bapat
EntepriseDB Corporation
The Enterprise Postgres Company