[mythtv-users] Client-Server hardware arch questions?

[Max]

Hello, I am a Myth newbie so bear with me if some of the questions are silly.

I am a 3+ year TiVo user and recently me and a friend of mine decided to play with the free linux pvrs out there. My friend picked freevo and I went with MythTV. The reason for my choice is the client server architecture - the potential of it just awed me. This is exactly the type of thing I always wanted.

So I wanted to understand how this works and a good way to put together a hardware platform to make this most usefull. What I am seeing in ideal world is this:

A heavy use server(or several) with PVR250 (or several) and a lot of hard drive space (maybe USB2/Firewire for expanding). The server will run backend part only, and maybe not even have a sound card.

Several lightweight frontend machines which are pretty much set-tops:
        * Diskless (they can boot off the server or flash)
        * Fanless (or as close to silent as possible)
        * Small
        * Good Quality Video Out (Composite/S-Video/maybe also a VGA-1080i converter for my Wega)
        * Good Quality Audio Out (Option for optical out?)
        * Contain DVD/CD drive for watching DVD/*VCD's, listening/ripping music CD's, Listening to MP3-CD's.

So here are the questions:

* How does the Myth frontend talk to backend? Is the video (audio, still picture, etc) transmitted in-band or do I need to NFS export the directory containing the common files? Is the media stream using TCP or UDP?

* What are the bandwidth requirements for connecting FE to BE? My calculations using 2GB per hour as top end PVR250 recording puts it at about 5 Mbit/s sustained w/o overhead(which pretty much excludes 802.11b, but a/g or FastEthernet may work) Are these numbers anywhere near reality?

* I assume the decoding is done on the front end, are there any hardware MPEG accellerators that can be used to offload decoding on the front end? I mean something a bit more lightweight than PVR250 (which is what I have now) - I only need to decode, not encode. Would the old DVD MPEG cards work? Any of them supported under linux?

* When you have multiple backends, do they communicate to determine optimal recording capabilities? Can you set priorities for recuring recordings to auto-determine what is recorded in case of conflict?

* When you have multiple backends, are they seen as one by the frontend?

* Where do the "plugins" run? On the backend? frontend? both?

* How many frontends can a backend support?

* What is the best cost effective frontend (as described above) hardware?

* Is there a possibility of (imaginary right now) "MythDVD" plugin allowing DVD playback running on frontend only?

* Am I asking way too many questions?

-Max-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/attachments/20030513/89dffb6c/attachment.htm