Re: SQUID I/O

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Sat, 06 Dec 1997 01:00:23 +0100

Alex Rousskov wrote:

> I do not measure the proxy->client latency. I measure the time it takes
> _Squid_ to get rid of an object. I understand that an object could be
> buffered on the way through the TCP stack. It does not look like it is
> just _copying_ though: the delays are much bigger.

It is expected that requests smaller than the socket buffer gets
delivered almost immediately in Squids point of view, since squid only
delivers the request to the buffer and closes the socket. It is then up
to the TCP/IP stack to deliver the pending data to the client. Requests
larger than the socket buffer should begin to be affected by network
delays.

Don't forget to account for the disk delays when opening and reading the
request from disk. And on a busy server things go quite a bit slower
since Squid only processes 4KB at a time. Between every 4KB block it
checks for new connections several times, and transfers data on other
pending connections.

---
Henrik Nordström
Received on Fri Dec 05 1997 - 16:23:32 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:51 MST