Round robin with Zope?

[zellster]

Hello,

I am planning to run multiple instances of Zope as an application server. There will be 1 or more Zope web servers running on back-end machines being served through a front-end web server.

nginx can distribute requests via weighted round-robin given a configuration like the following:

upstream zope {
server 192.168.123.45:8080;
server 192.168.123.45:8081;
server 192.168.123.46:8080;
server 192.168.123.46:8081;
}

Can LSWS do something similar, or would I need LSLB?

Thanks.

[mistwang]

You can do the same with LSWS, the load balancing algorithm is "Least Load".
LSLB can give you more flexibilities on load balancing algorithms and better reliability in handling of backend failures.

[mistwang]

FYI: Your email server bounced the forum notification email with "No such user here".

[zellster]

Originally Posted by mistwang

You can do the same with LSWS, the load balancing algorithm is "Least Load".
LSLB can give you more flexibilities on load balancing algorithms and better reliability in handling of backend failures.

I appreciate the information. Other questions for LSWS:

* Where in the documentation would I look for examples
* Are connections to the back-end persistent
* How are back-end failures handled

Thanks again.

[mistwang]

We do not have an example dedicate for LSWS load balancer configuration, it should be easy to configure with our web console with online help.
you can search for examples using LiteSpeed with mongrel cluster.

The connection is default to persistent and configurable.
LSWS will try other back-end nodes whenever possible if one node fail. The request will take longer time to process, if all nodes are bad, 503 will be returned.