Re: [squid-users] Performance tips for accelerator setup

From: Matus UHLAR - fantomas <uhlar_at_fantomas.sk>
Date: Tue, 26 Jan 2010 10:27:39 +0100

>> On 16.01.10 14:35, Amos Jeffries wrote:
>>> Also, you may want to use CARP from the parents to siblings. That
>>> will restrict the objects each sibling is caching to the URLs CARP
>>> assigns it. Less duplicated objects leading to a longer caching time
>>> window and more hits overall.

> Matus UHLAR - fantomas wrote:
>> Is this usable between peers manually configured or does it require support
>> from network layer? I've always thought it requires intercepting and
>> encapsulating HTTP connections by intermediate routers...

On 26.01.10 22:23, Amos Jeffries wrote:
> No. It's just a way of selecting parent caches. Just like round-robin or
> source-hash. One non-storing gateway instance with multiple backend
> caching instances.

I apparently have mistaken it for WCCP :)
but it still requires multiple levels which is currently no way for us.

>>> If you are using ICP between the siblings, try HTCP for better peer HITS.
>>
>> last time I asked about HTCP, I got answer that there's no benefit from it
>> yet. When did this change?

> How long ago was last time? since 2.7 came out the CLR and several
> update bits have been added that speed up obsoletion and
> if-not-modified. It might just be opinions too.

does it apply for 3.0 too? Or at least for 3.1 (which is afaik
soon-to-be-released?)

-- 
Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
Posli tento mail 100 svojim znamim - nech vidia aky si idiot
Send this email to 100 your friends - let them see what an idiot you are
Received on Tue Jan 26 2010 - 09:27:50 MST

This archive was generated by hypermail 2.2.0 : Tue Jan 26 2010 - 12:00:04 MST