Subject | Re: [ib-support] Re: no gds_inet_server on Red Hat 6.2 |
---|---|
Author | Paul Reeves |
Post date | 2001-11-30T13:59:50Z |
henryfranquet@... wrote:
lock resolution will also be quicker. Under CS each user has their own cache
and the lock file must be constantly updated by each user (ie, each process).
If/when things go pear shaped all super server processes will die and all
uncommitted work is lost. Under CS only the affected process dies. Everyone
else continues quite happily.
to set tcp_no_delay (which is the option I think you are referring to.)
However, I do not believe that anyone has properly tested it. (ie, the switch
will work but we are not sure if it yields any benefit.)
If anyone has played with this option on FB1 RC1 I'd be interested to know.
Paul
--
Paul Reeves
http://www.ibphoenix.com
taking InterBase further
>Classic is slower, but more stable. SS allows all users to share the cache and
> I haven't yet test Firebird with my app. But is Classic version more
> stable or faster than SuperServer Version ? I have very heavy query
> application ( I shall use GIGABIT connexion) and rather heavy OLTP
>
lock resolution will also be quicker. Under CS each user has their own cache
and the lock file must be constantly updated by each user (ie, each process).
If/when things go pear shaped all super server processes will die and all
uncommitted work is lost. Under CS only the affected process dies. Everyone
else continues quite happily.
> Thanks, but my first objective is : Is there a optimation in TCPIt doesn't exist for the InterBase SuperServer release. Firebird has an option
> protocol for SuperServer Version, as to avoid wait state ?
>
to set tcp_no_delay (which is the option I think you are referring to.)
However, I do not believe that anyone has properly tested it. (ie, the switch
will work but we are not sure if it yields any benefit.)
If anyone has played with this option on FB1 RC1 I'd be interested to know.
Paul
--
Paul Reeves
http://www.ibphoenix.com
taking InterBase further