On Mon, Sep 28, 2009 at 7:19 PM, Stuart Beasley-Forrest <
thefirststarea...@gmail.com> wrote:

> Our cataloger reports that when searching OCLC via Z39.50 the results take
> at least 30 seconds to return. I have myself tried this at home and the time
> period is roughly the same (within a few seconds). Does the search from the
> staff client go to the evergreen server first then to OCLC or is it direct
> to OCLC from the staff client.
>

The former.  Since you do not complain about the performance of the staff
client overall, one might expect a problem between the EG and OCLC servers.
While it might seem like it would be more efficient to query directly from
client to OCLC, there are two opposing considerations:

   - coding a z39 client in javascript/XUL is huge undertaking, and probably
   a fruitless one because
   - once you start having to push all the files back UP to the server, you
   lose any of the efficiency... and potentially more, due to asymmetric
   bandwidth limits.

Ideally, you should do some network analysis on the server-side connection,
perhaps by replicating the same search performed via staff client on the
server command-line using yaz-client.  You should also ask your network
administrator about any QoS or port-based restrictions that would affect the
OCLC connection.  It is fairly common to restrict bandwidth on the less
standard ports.

You might also try to search the same target using a similar interface, like
the one in ‡biblios, for comparison.

--Joe Atzberger

Reply via email to