Hello.
Bryan Pendleton wrote:
If overall system load doesn't drop much at all, then would the
conclusion be that the patch makes the server better, but the client
has actually become worse due to using a less efficient algorithm
to handle the no-length-header message format?
To result the patch makes server better in streaming,
it is needed to measure the load of server in streaming separatedly from
server and client.
However I think it is impossible because streaming is process of
synchronized pair of server and client.
Load of only one side in streaming have no meaning, I think ...
Fact, I think, is that suspicious matter which may caused the problem of
the patch was just founded at client side.
Here, suspicious matter is that current client implementation may not
works better without length of streamed information than with it,
because client does not know the size of information to be buffered.
And there are a circumstantial evidence for my suspect that
performance of streaming blob was improved after the patch
with modification of the size of buffer at client side to the size of
binary data to be streamed.
Well ....
Assuming suspicious matter, I will think some countermeasures.
We need them.
Best regards.
--
/*
Tomohito Nakayama
[EMAIL PROTECTED]
[EMAIL PROTECTED]
[EMAIL PROTECTED]
Naka
http://www5.ocn.ne.jp/~tomohito/TopPage.html
*/
