Ken,

Thanks for the insights.  Are you describing this scenario:

- Load balancer be put in the first tier
- The application and search engine be on the second tier.
- The load balancer communicates to to the second tier using PB and
using only one connection per process can still send numerous
messages.

What would you suggest to use as a load balancer, since it would need
PB support?

On 6/30/06, Ken Kinder <[EMAIL PROTECTED]> wrote:
On 6/29/06, joe kim <[EMAIL PROTECTED]> wrote:

> I need to do more research on PB.  Are you saying that PB can function
> as a HTTP load balancer?


No. The use-case I have in mind is that you already have an application with
an existing load balancer. When you add full-text indexing to your
application, you have another server setup with nxlucene. The web servers
query that server for data.

Of course, xmlrpc is one way to query them, but it has some drawbacks. One
connection can only handle on concurrent request, and with HTTP/1.0, only
one request. With pb, besides some improvements in the binary nature of the
protocol, you can handle hundreds (thousands?) of queries with one
connection. So having one permanent connection instead of thousands of
transient connections gives you some scalability advantages.

-Ken



_______________________________________________
pylucene-dev mailing list
[email protected]
http://lists.osafoundation.org/mailman/listinfo/pylucene-dev



_______________________________________________
pylucene-dev mailing list
[email protected]
http://lists.osafoundation.org/mailman/listinfo/pylucene-dev

Reply via email to