tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Enrique Arizón <e_arizon_ben...@yahoo.com>
Subject Re: Tomcat clustering
Date Mon, 07 Apr 2008 21:28:36 GMT

--- Christopher Schultz <chris@christopherschultz.net>
escribió:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Enrique,
> 
> Enrique Arizón wrote:
> | If you users sessions are simple tomcat sessions
> clustering will be
> | enough but if you have complex session data
> structures (let's say for
> | example a java.util.ArrayList/Hashtable) you will
> need custom
> | solutions to sync sessions between clustered
> nodes.
> 
> Really? My experience has been that anything that
> properly implements
> java.io.Serializable does just fine migrating from
> node to node. Sure,
> migrating lots of data around a cluster is
> time-consuming and perhaps
> would not be considered a best-practice, but the
> container-provided
> clustering and session sync capabilities certainly
> do work.

 Completly agree. The case is that serialization is
slow, really, really slow (at least in terms of CPU
cicles) even serializing to local disks. But no doubt
it works. What I mean is that trusting container
clustering is going to solve any clustering problem
can be a really wrong idea, mainly because the
container limits to (correct me if I'm wrong)
serialize over the network session data. If such
session data is just "a pointer" to real data, it can
work, otherwise it will miserable fail with no so big
loads.

 On the opposite clustered databases will limit to 
sync changes (maybe just a few rows amongs maybe
thousands of rows per user). Also a custom solution (a
node notifying it's peers "I changed this and this
, please update") will work much better.

> 
> I would argue that a small session footprint is
> critical to a robust web
> application, clustered or not.
> 
> | Mysql makes a great help since it use in memory
> (RAM) storage for
> | clustered tables
> 
> What are you talking about? The closest thing to
> "clustered tables" that
> MySQL supports is the FEDERATED storage engine, and
> there are no
> guarantees about RAM usage for it.

Guess we are not speaking about the same thing. I
refer to NDB Cluster engine:
 
http://dev.mysql.com/doc/refman/5.0/en/mysql-cluster.html

 I read time ago that Google improved such engine for
its own purposes and the next Mysql will show improved
clustering stuff. Also, since Mysql is now part of Sun
I guess also it will get a boost from Solaris
engineering (I mean, Dtrace).


> 
> - -chris
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.8 (MingW32)
> Comment: Using GnuPG with Mozilla -
> http://enigmail.mozdev.org
> 
>
iEYEARECAAYFAkf6gT0ACgkQ9CaO5/Lv0PAV2wCgtSxf1Pcwi/8cXEuKsnu4EPAe
> 9hoAnir3tkn+qZ6l0pnoXhf1r4c9gVfY
> =QyZ9
> -----END PGP SIGNATURE-----
> 
>
---------------------------------------------------------------------
> To start a new topic, e-mail:
> users@tomcat.apache.org
> To unsubscribe, e-mail:
> users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail:
> users-help@tomcat.apache.org
> 
> 


---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org


Mime
View raw message