[multicat-devel] Multicat reads and HDD disk I/O

[david miller]

Just chiming in, and I'm sure others will as well.

Most IP/UDP video is TS-based and therefore latency sensitive.  You want to spit it out as fast as possible, and buffer on the receiving end with the CPB.  If you have disk I/O issues, you need a faster disk, IMHO.  I can't think of a scenario where I would ever want to buffer like this <ducks>.

FWIW, I would try a commercial streaming solution.  Multicat is good for testing/recording/diagnosis purposes, but is only as good as it's implementation.  If you are having problems with disk I/O, this may not be the solution for you.  There's many other I/O issues in trying this as well (NICs and bus hardware/etc come to mind).

VideoLAN may provide some commercial support if you ask nicely :-).

-Mike
--------------------------------------------
On Mon, 9/29/14, Ibrahim Tachijian <barhom at gmail.com> wrote:

 Subject: Re: [multicat-devel] Multicat reads and HDD disk I/O
 To: "Mailing list for multicat developers" <multicat-devel at videolan.org>
 Received: Monday, September 29, 2014, 3:35 PM
 
 Sorry I
 meant reading from file to output to multicast.
 On Mon, Sep 29, 2014 at
 9:28 PM, Ibrahim Tachijian <barhom at gmail.com>
 wrote:
 Hello,
 We are
 running a system where we launch and stop multicats via
 scripts and are having some I/O issues with multicat read
 operations (read from file output to disk).
 Our greatest concern is that the
 program does not buffer any kind of data on the reads and
 thus generates too much I/O when streaming. Are our concerns
 correct? Does introducing a read buffer to multicat
 (multicat would have greater memory usage) help in the
 situation ?
 Regards,
 -- 
 Ibrahim Tachijian
 
 
 
 
 -- 
 Ibrahim Tachijian
 
 -----Inline Attachment Follows-----
 
 _______________________________________________
 multicat-devel mailing list
 multicat-devel at videolan.org
 https://mailman.videolan.org/listinfo/multicat-devel