Re: I have read and worked since 2 weeks ago

From: Amos Jeffries <squid3_at_treenet.co.nz>
Date: Tue, 19 Jul 2011 12:14:49 +1200

 On Mon, 18 Jul 2011 17:20:52 +0430, Mohsen Pahlevanzadeh wrote:
> Dear Henrik,
>
> I'm just a programmer and not network architecture, However i know
> network just about TCP/IP and some of protocols, Also already i
> design a
> protocol and implement it.Now, network architectures of our company
> design a big plan and i had implement 10% of it. My implementation
> need
> to know data follow of squid, before it, i was working on squid 2.7
> and
> Amos suggested i work on 3.x, now i was working on 3.1.14, I read the
> following paper :
> http://www.squid-cache.org/Devel/papers/chadd-clientlet.txt
> But, it's very old, and it explained in theory.
> I need to theory + code.

 That appears to be Adrians personal TODO list with designs he wanted to
 add to squid 2.7 but never got around to.

 The clientlet and serverlet he is describing are 100% internal
 components of Squid (and fictional). Even if Squid was implemented that
 way the direction of of HTTP traffic flow is _directly opposite_ to to
 the protocol data flow you have decided to use.

 Being personal notes the text uses the word "client" very loosely and
 sometimes it means an application, sometimes a function call, sometimes
 a component in the abstract design. Even so, you can see there is still
 no mention of a server being given _response_ by a client. It is all
 about requests.

 Amos

> You made me happy (if possible ) talk about squid data follow [squid
> architecture] !
> At finally, sorry for my english language.
>
> Yours,
> Mohsen
> On Mon, 2011-07-18 at 13:17 +0200, Henrik Nordström wrote:
>> mån 2011-07-18 klockan 12:42 +0430 skrev Mohsen Pahlevanzadeh:
>>
>> > > a) How do I fetch content via Squid?
>> > I don't want to squid fetch pages, But also i have 2 machine A and
>> B.B
>> > is squid-box and A get pages from Internet and send to squid.Squid
>> > "must" cache them and doesn't any response.
>>
>> What is the difference?
>>
>> Are you trying to solve an actual problem in the bast manner
>> possible,
>> or are implementing a fixed design even if there is simpler and much
>> more maintainable ways of solving the same problem?
>>
>> What I am trying is to explain to you how you most easily solve the
>> problem you described (how to push arbitrary content into the Squid
>> cache), while keping full control of how the content enters the
>> squid
>> cache.
>>
>> The Squid cache is an HTTP cache, and the most convenient way of
>> entering content into the cache is by using HTTP.
>>
>> What you need for this to work is to write a small HTTP client
>> reading
>> the list of URLs and make your 'A' box implement a small HTTP server
>> allowing Squid to fetch the content from there when requested by
>> your
>> HTTP client. This effectively pushes the content into the Squid
>> cache
>> and requires no modifications to any version of Squid.
>>
>> I have implemented exactly this scheme in the past for a satellite
>> provider pushing a stream of HTTP objects over a one-way satellite
>> link
>> and it worked out quite nicely.
>>
>> Yes there is some extra overhead involved as Squid do forward the
>> response to your http client while it is cached, but thats just
>> local
>> either within the squid server bix or between the Squid server and
>> the
>> content pushing server depending on where you locate the http client
>> component.
>>
>> Regards
>> Henrik
>>
Received on Tue Jul 19 2011 - 00:14:54 MDT

This archive was generated by hypermail 2.2.0 : Tue Jul 19 2011 - 12:00:03 MDT