Mailing-List: contact dev-help@tomcat.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Tomcat Developers List" <dev@tomcat.apache.org>
Received-SPF: pass (athena.apache.org: local policy)
Message-ID: <3ca3a8bc45c7e2868f3cedba1ed897cf.squirrel@mail.mohawksoft.com>
In-Reply-To: <353769790912171143u26895067x1c2db2eea66bc95b@mail.gmail.com>
References: <353769790912111837u6a1e76fana32028ff6f301550@mail.gmail.com>
    <4B239567.9000403@mohawksoft.com>
    <353769790912171143u26895067x1c2db2eea66bc95b@mail.gmail.com>
Date: Thu, 17 Dec 2009 15:24:35 -0500
Subject: Re: Storing Tomcat Sessions to MySQL (and Memcached) Round-Robin  
 without sticky sessions
From: markw@mohawksoft.com
To: "Tomcat Developers List" <dev@tomcat.apache.org>
User-Agent: SquirrelMail/1.4.19
MIME-Version: 1.0
Content-Type: text/plain;charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Importance: Normal

> Hi Mark, Thanks for the reply.
>
> In your opinion should we be using local memory to store any session
> data? The discussion right now is to store everything to cache+db and
> nothing in local memory.

The problem with any cache system is cache coherency. Tomcat takes the
easy way out by requiring that sessions be "sticky" to a single box, and
using local access management. Sticky session have well known scalability
issues.

The "right" answer, for an arbitrary definition of "right," is a little
more complicated.

A UDP based server to reduce the TCP/IP overhead.
The protocol should be something like this: (pseudo code)

class Session
{
    UUID session;

}
Object lockSession(uuid session)
Object lockSession(Object session)
void unlockSession(Object session)
void abandonSession(uuid session)
Object peekSession(uuid session)
Object peekSession(Object session)


> On Sat, Dec 12, 2009 at 5:06 AM, Mark Woodward <markw@mohawksoft.com>
> wrote:
>> About 12 years ago, I wrote a system called msession for PHP that was
>> basically a high-speed cache for sessions. I have been using it for PHP
>> sites ever since. The one thing I've never liked about tomcat is sticky
>> sessions. It doesn't scale well enough. With gig ethernet, a shared
>> network
>> based caching system is probably the best way to go.
>>
>> Also, it would be good if the format of the session data could be used
>> by
>> other platforms, i.e. PHP and Tomcat could share sessions.
>>
>> Is anyone actively working on the session management of tomcat?
>>
>>
>> anthonyvierra@gmail.com wrote:
>>>
>>> Hello, Thanks in advance for your help.
>>>
>>> I would like to know if anyone is currently modifying, has modified,
>>> or plans to modify Tomcat 6.0.20 to store/read sessions directly in
>>> Memcached + MySQL to enable session persistence with round-robin and
>>> without sticky sessions?
>>>
>>> If so would you be so kind as to point me in the right direction?
>>>
>>> Our top level design requirement is 'no sticky sessions' + 'round
>>> robin' + 'MySQL or Amazon SimpleDB'
>>>
>>> After getting this solution working I would like to donate the code to
>>> the Tomcat project. It will be useful for those wishing to deploy
>>> Tomcat + Amazon EC2 + Amazon Elastic Load Balancer.
>>>
>>> Thank you,
>>> Anthony
>>> 925-456-4343
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
>>> For additional commands, e-mail: dev-help@tomcat.apache.org
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: dev-help@tomcat.apache.org
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: dev-help@tomcat.apache.org
>


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org