Same Apache server process?

I have a dumb question:

We all know Apache starts up a number of processes for load balancing.
When a user connects to Apache and browse through web pages, does the
Apache server process that responded to the user request first keep
replying to the user, or do different Apache server processes reply to
the user as the user browse through web pages? In other words, do
users get dedicated Apache server for them? Thank you in advance.

1443131@usenetplanet.com said:
>We all know Apache starts up a number of processes for load balancing.

Not actually for load balancing, but to be able to faster serve the
client requests. If the child processes were created only after
the user connections, there would be an additional delay equivalent
to the time needed to start the handler process in serving all the
requests. Thus, Apache is reducing the service time delay by starting
some server processes beforehand.

>When a user connects to Apache and browse through web pages, does the
>Apache server process that responded to the user request first keep
>replying to the user, or do different Apache server processes reply to
>the user as the user browse through web pages?

This is a different issue, and it depends.

In any case, when a TCP connection request to the server arrives, it is
delegated to one of the available server processes (I understand Apache
maintains an ordered list of processes, and new connection is delegated
to the lowest-numbered/"first" available process in this list).

In HTTP/1.0 and earlier, only one page element was served per TCP
connection (so, element as in a picture, or stylesheet, or HTML
document, etc). So, a typical page load would consist of a multiple TCP
connections, and thus most possibly the client would end up being served
by multiple Apache processes even in the context of loading a single
"page". Depending on a client software, it would even be possible that
the client was at any given time having multiple requests open to the
server (f.ex. one for the main HTML document, and few more to retrieve
the various images on the page).

With HTTP/1.1 came the concept of keep-alive sessions - though even
now it is possible for the client to request a non-keepalive session,
and also it is possible for the server to refuse a request for a
keepalive session from the client, and only serve in the HTTP/1.0-like
non-keepalive mode. With keepalive sessions, the client can request an
arbitary (or server-limited) number of HTTP resources using a single
TCP connection (and thus just a single Apache server process). There is
a server-imposed timeout in the keepalive sessions, so if there is no
traffic between client and server for some given amount of time, the
server will terminate the connection. The client can also terminate the
connection, f.ex. if the user directs the client to a page residing on
some other server.

>In other words, do users get dedicated Apache server for them?

So, users may get a server process that is dedicated for some span of time,
but this is nowhere near guaranteed, and this happens mostly to improve the
efficiency at the TCP connection level (to avoid multiple TCP connections
for serving just a single page).
--
Wolf a.k.a. Juha Laiho Espoo, Finland
(GC 3.0) GIT d- s+: a C++ ULSH++++$ P++@ L+++ E- W+$@ N++ !K w !O !M V
PS(+) PE Y+ PGP(+) t- 5 !X R !tv b+ !DI D G e+ h---- r+++ y++++
"...cancel my subscription to the resurrection!" (Jim Morrison)