Path: csiph.com!usenet.pasdenom.info!aioe.org!news.stack.nl!newsfeed.xs4all.nl!newsfeed3.news.xs4all.nl!xs4all!newsgate.cistron.nl!newsgate.news.xs4all.nl!post.news.xs4all.nl!not-for-mail
MIME-Version: 1.0
In-Reply-To: <0626cc7c-d4fd-4d6e-ac3d-9a85a725b5fb@googlegroups.com>
References: <0626cc7c-d4fd-4d6e-ac3d-9a85a725b5fb@googlegroups.com>
Date: Fri, 25 Oct 2013 21:25:16 +1100
Subject: Re: How to design:Use One TCPIP Client connection to Server and work concurrently
From: Chris Angelico <rosuav@gmail.com>
To: python-list@python.org
Content-Type: text/plain; charset=ISO-8859-1
Precedence: list
Newsgroups: comp.lang.python
Message-ID: <mailman.1505.1382696725.18130.python-list@python.org>
Lines: 43
NNTP-Posting-Host: 2001:888:2000:d::a6
Xref: csiph.com comp.lang.python:57511

On Fri, Oct 25, 2013 at 9:06 PM, ray <yehrayyeh@gmail.com> wrote:
> Hi all,
>   The system use ONE tcpip client and connection to server.
> A large batch transactions send to server.I want to send to server concurrently.
> any idea ?
> In my thinking,
>  A.Devide large batch transactions to small batch transactions.
>  B.Create multi-thread or multiprocessing to process small transactions.
>  C.No idea at how to control multithread using  ONE client concurrently?
>    Each small transactions do not need to wait others
> Thank for your suggestions.

Are you able to open multiple TCP/IP connections to the server and run
them in parallel? If not, you're going to be limited to what the
server can do for you. Otherwise, though, it's just a matter of
working out how best to manage multiple connections.

* You can manage everything with a single thread and asynchronous I/O.
This would probably be easier if you're on Python 3.4, but it's
possible with any version.

* Or you can spin off threads, one for each connection. This can work
nicely, but if you don't know how to get your head around threads, you
may confuse yourself, as has been recently discussed in other threads
(discussion threads, that is, not threads of execution - what a
crackjaw language this English is!).

* Or you can use the multiprocessing module and divide the work into
several processes. Again, you have to figure out what you're doing
where, but this can in some ways be a lot easier than threading
because each process, once started, is completely separate from the
others.

* Or, rather than manage it within Python, you could simply start
multiple independent processes. Somehow you need to figure out how to
divide the transactions between them. Could be really easy, but could
be quite tricky. I suspect the latter, in this case.

There are many ways to do things; figuring out which is the One
Obvious Way demands knowledge of what you're trying to achieve. Do any
of the above options sound good to you?

ChrisA