mickael.remond at process-one.net
Fri Sep 15 12:49:33 MSD 2006
* Simon Yeo <simon at meebo.com> [2006-09-14 23:13:01 -0700]:
> I guess my question is if you folks have seen any evidence that having
> shared heap is helping any. Does it definitely improve ejabberd's
> memory usage problem? Does it impact performance one way or another?
As I said, nor -shared nor -hybrid will help in this case. It will
make things worst.
>From what I know, I think your usage pattern is a bit special, as your
memory consumption is not at all usual for your level of usage.
How to solve it really, really depends on the detail of your
requirements, your constraints, pattern usage, etc.
We are working on several new approach that are being experimented in
the "lab" and that make ejabberd even more scalable.
Normally, with the current version, in a cluster, you should be able to
support at least 150k users.
We are working currently working to break this limit by a large margin
:-) Some of the improvement will probably be ready for the 1.2 version
(The next one will be 1.1.2).
For large sites, you must be aware that you cannot take ejabberd as a
product. At some point, all IM server should be used as tool.
Running a service with large amount of users is _HARD_. _Very hard_.
You cannot take a software out of the box deploy it and connects
millions of users. Never. This happens in big hardware vendors
commercials, but not in real life. You have to know the context of your
deploiement. You have to know your tool very well. You have to change
it, improve it, optimize it.
More information about the ejabberd