[mythtv-users] Anyone running a diskless frontend?

[usenet at wingert.org]

Personally, I would not run X over ethernet.  I would export the root of a
filesystem and run X locally on the diskless machine (much less bandwith 
given the raw bulk of video).  Although I export mythtv sessions all 
the time from one of PVRs for testing purposes, performance is fine.

If you are exporting the streams as opposed to X, the following should 
yield your performance.

I haven't had much experience with the codec that MythTV uses (I am using TiVos for recording devices).  If I remember correctly 1 hour is around 2 GBs.  
This is about 582 kB/s.

A good 100BT connection can sustain about 90 Mb/s (11 MB/s), in an optimal 
environment.

Given this performance 100BT should be able to sustain approximately 19
streams.  This does not include the disk bottleneck though.

Since X exports images in a more raw format, you essentially lose your
codec compression rate.  So you could divide the number of streams by
your compression rate to achive the number of X streams.

BTW - 100BT has a lot of bandwidth.

HTH

> 
> Further to this discussion, I have been thinking about this.  <Disclaimer>  
> this is a question, not a suggestion.  I haven't tried it and doubt it will 
> be very efficient.</Disclaimer>
> 
> How about running the backend *and* frontend somewhere in a storage room or 
> whereever you can't hear it.  Then next to your TV you have a Pentium 200 or 
> some such just running X -query mythserver?  Vanilla pentiums used to be 
> quiter than a mouse.  Double that with a nfs-root and you have a dead silent 
> "frontend" that only has a network card and a TV-out PCI card.  The video 
> card of course must have xv support.  You have to take care of audio.  How 
> about nas?
> 
> Would the traffic choke a 100MB switch?  Is this even feasable to attempt?
> 
> Thanks,
> IvanK.
> 
> On Thursday 13 February 2003 04:09 am, Bruce Markey wrote:
> > Aaron Stewart wrote:
> > > I dunno.. It seems like network transfer of data is adding another level
> > > of complexity that could be avoided, but if sound dampening is
> > > necessary, then it's a necessary evil :).
> > >
> > > My understanding was that an uncompressed mpeg2 stream ran at
> > > 18mbits/sec, which translates to:
> > >
> > > send->18mbit
> > > recv<-18mbit
> > > buffer->18mbit (for delayed playback)
> 
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at snowman.net
> http://www.snowman.net/mailman/listinfo/mythtv-users
> 
>