Re: Performance benchmarking BOF

From: Alex Rousskov <rousskov_at_measurement-factory.com>
Date: Tue, 24 Oct 2000 10:13:17 -0600 (MDT)

On Tue, 24 Oct 2000, Brent Mills wrote:

> I came across an interesting thread regarding streaming media
> benchmarking and the difficulty in producing a good tool. This
> thread was from 5/2000. Do you know if any progress has been made
> in this area?

We have been talking to Microsoft folks to get access to their
protocol internals. At this time, their recommendation was to use
various Windows Media SDKs in order to build the benchmark. Based on
other developers comments, I am somewhat skeptical that SDKs will be
sufficient. However, we must actually try to use them to know for
sure. So far, no code was written; we were reading SDK and protocol
documentation. We plan to start writing the code within a few weeks
from now.

A few vendors stand behind and are eagerly waiting for the results of
this development, which is a very good thing.

We have not talked to Real folks about getting access to their
protocol. However, based on general experience with Real, they are
even more protective about the protocol details and worse on developer
support than Microsoft is. The former is probably inevitable when you
have to compete with a rich and "active" company like Microsoft.

My understanding is that the idea of implementing the benchmark based
on Quick Time was killed by the pessimism of most interested parties.
Nobody seem to care about Quick Time enough.

Finally, the idea of designing a *protocol* that is neither Real, nor
WinMedia, nor QuickTime, but is sufficiently close to the popular
protocols is still alive. However, we need to gain more knowledge
about the trade-offs before we can decide on that one. The work
described above should give us that knowledge.

Thanks,

Alex.
Received on Tue Oct 24 2000 - 10:13:24 MDT

This archive was generated by hypermail 2.2.0 : Thu Nov 20 2008 - 12:00:03 MST