Re: The problem is that SQL*Net is too chatty, because FTP runs fine.

Rajeev,

Platforms and SAN are the same in all data centers. Ping and traceroute are blocked by the firewalls so we are unable to gather any network statistics. My understanding is that the network group has collected stats, but they are have only provided their summary analysis and not the actual statistics nor their testing methodology. We know the number of hops and firewalls on the southwest connection. We have no information how many hops or firewalls there are on the southeast connection. We do know that the distance to the southwest data center is more than the distance to the southeast. Of course depending on how the routes are set up, it is entirely possible that the path to the southeast is actually longer. We also know that the same refresh takes 15 times longer on the southeast pipe than it does on the southwest pipe.

David

On Jan 31, 2008 1:19 PM, Rajeev Prabhakar <rprabha01_at_gmail.com> wrote:

>
> From a pure network standpoint, perhaps, you could collect network
> latency / packet loss (if any), RTT figures between the relevant sites
> and if the figures are vastly different OR are not making sense, share
> it with you network folks.
>
> i.e. The ones showing acceptable data transfer time AND
> The ones showing unacceptable data transfer time.
>
> Bottomline - You got to talk their language (i.e network statistics).
>
> Are these sites same in terms of software/hardware/storage layouts ?
>
>
>

--
http://www.freelists.org/webpage/oracle-l