tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Peng Tuck Kwok" <pengt...@gmail.com>
Subject Re: Tomcat Performance
Date Wed, 12 Nov 2008 14:21:17 GMT
David, was replying to Charles's earlier email on that.
I thought I read it right first in the email and assumed Charles was correct
in that.
Hazards of reading mail on the go :P


On Wed, Nov 12, 2008 at 9:26 PM, David kerber <dckerber@verizon.net> wrote:

> Peng Tuck Kwok wrote:
>
>> Probably the reason why he's seeing one instance of tomcat moving quicker
>> than 2 instances is the fact that there is some form of contention for
>> resources on that single machine assuming that the 2 instances are
>> configured identically in every aspect (other than ports).
>>
>>
> You mis-read it.  He's seeing twice the performance from two instances than
> he is one single one, which shouldn't be if the one is properly configured,
> (with the exception noted below).
>
> D
>
>
>  The idea is not to give you a 0-60 mph capability with 2 tomcats on a
>> single
>> box (partition) but to give you better throughput. As I understand it,
>> when
>> you start getting more load, you'd be able to handle the requests in a
>> linear fashion (again assuming you've sized the 2 or more instances
>> correctly).
>>
>> *>I would rarely recommend that a client run parallel app servers on the
>>
>>
>>> same machine for the same application for any purposes other than being
>>> able to switch between versions of the same application (say, for
>>> zero-downtime upgrade strategy).
>>>
>>>
>> *I wouldn't recommend anyone do that just to switch versions for a zero
>> downtime upgrade strategy as well. Some sort of DR would be better for
>> this
>> ? Down production and switch to DR then when upgrades are complete just
>> reverse what has been done.
>>
>> *>Since the OP didn't say that's what his
>>
>>
>>> requirements were, there doesn't seem to be a compelling reason to use
>>> this strategy.
>>>
>>>
>> *You're right, until we really know what his requirements/KPI's on that
>> are
>> then most of this is largely academic.
>>
>> Nishi, the link to the redbook is here
>> http://www.redbooks.ibm.com/abstracts/sg246392.html?Open .
>> It's websphere specific, but there's still lot of things you can pick up
>> on
>> and probably apply.
>>
>> On Wed, Nov 12, 2008 at 5:15 AM, Christopher Schultz <
>> chris@christopherschultz.net> wrote:
>>
>>
>>
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> Pengtuck,
>>>
>>> pengtuck@gmail.com wrote:
>>>
>>>
>>>> So let me get this straight. You are reluctant to accept a
>>>> configuration which gives you improved throughput ? :P
>>>>
>>>>
>>> No, the OP is unwilling to use a configuration that doesn't make any
>>> sense: one single Tomcat should outperform two Tomcats on the same
>>> physical server (unless you are talking about a 32-bit JVM that needs a
>>> lot of memory).
>>>
>>>
>>>
>>>> Anyway, this is not an unusual approach, from what I understand this
>>>> simply makes full use of the resources available on that machine. Not
>>>> uncommon in real world to see app servers like websphere being
>>>> configured in that manner.
>>>>
>>>>
>>> I would rarely recommend that a client run parallel app servers on the
>>> same machine for the same application for any purposes other than being
>>> able to switch between versions of the same application (say, for
>>> zero-downtime upgrade strategy). Since the OP didn't say that's what his
>>> requirements were, there doesn't seem to be a compelling reason to use
>>> this strategy.
>>>
>>> - -chris
>>>
>>>
>>
>
>
> ---------------------------------------------------------------------
> 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
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message