tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Preston L. Bannister" <pres...@bannister.us>
Subject Re: Are Sticky Sessions really necessary?
Date Sat, 03 Nov 2007 20:03:46 GMT
On 11/3/07, Len Popp <len.popp@gmail.com> wrote:
>
> But isn't the purpose of session replication to allow different
> servers to handle the session? If not, what's it for?
>


This discussion touches on a bit of philosophy, and a set of design
decisions built into the code, and no longer relevant.  In this case it's
"sticky sessions", a notion that once made some sense, but no longer.  There
are really only two design points that make sense for sessions:  1) sessions
kept in-memory (or on-disk) within a single machine, and 2) sessions kept in
an external service (database or specialized store).  There are a bunch of
intermediate solutions that used to make some sense - before web-oriented
databases, before massive scaling, before gigabit local nets, before cheap
and insanely fast single-box systems.

All the time spent maintaining intermediate solutions removes focus.  Java
is not losing out to PHP (and the like) due to lack of focus on the present,
not inferior technology.

Not that I expect the argument to make any headway in this venue, but
someone has to raise the issue.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message