This seems like a very large load for just one server:

Network:

Let's first assume that you use the MP3 maximum bitrate of 320 kbit/s.
This is equivalent to 40.000 bytes per second. With all 200 clients
running, you have 8.000.000 bytes/s throughput, so that the network
interface is nearly saturated, if you do not choose to use a gigabit
interface, which would by a logical choice, since the backbone switch
should be gigabit as well. Regarding your overall investment, this is
not that much more expensive. Whatever, network throughput should be no
problem.

Harddisk:

While the throughput of 8 MByte/s does not seem much at first sight
with current harddisks running at 30 MByte/s at least (inner tracks),
you have to remember, it is 200 clients causing that load, each
probably playing a different song.

This means that with buffering at - say - 1 second, there would be 200
random seeks per second, each taking ~12ms for a current harddisk
including latency and read. This would take 2.4 seconds, so it would
not work. You have to make sure that buffering is at least 3 seconds,
which equals around 128 KBytes. With a Linux server, you can control
this read-ahead, regardless of application buffering.

However, with real-life assumptions, maybe it would work out just fine,
since you don't have to use maximum bitrate and in a hotel, you'd rarely
see more than half of the clients actually working.

Processing:

This does not mean there is no other K.O. criterium here, I don't know
if the processor load caused by slimserver may be to high - after all,
it's interpreted PERL...

Maybe someone with more than a few clients can give figures about their
CPU load factors?


-- 
meyergru
------------------------------------------------------------------------
meyergru's Profile: http://forums.slimdevices.com/member.php?userid=1980
View this thread: http://forums.slimdevices.com/showthread.php?t=18098

_______________________________________________
Discuss mailing list
[email protected]
http://lists.slimdevices.com/lists/listinfo/discuss

Reply via email to