>> Also im curious as to what kinda clustering issues are there when you
distribute it out on your farm. does your controller create a bottleneck
issue?

I'll let you know when I do. Single sign-on/fail-over/network transparency
have all been deemed crucial to our business, so we are taking our time
evaluating these techniques before deployment.

Clarification: When I wrote "1000+ concurrent users" I am obviously talking
about open connections and not business transactions. Even on high volume
sites, simultaneous transactions are lower in number than concurrent user
sessions. I am not sure how many simultaneous transactions occur given 1000+
concurrent users. I'm not saying that it doesn't happen, I just want to
clarify that concurrent users does not usually equate to simultaneous
database operations.

On 11/21/05, Patrick Casey <[EMAIL PROTECTED]> wrote:
>
>
> If you're not talking about a large number of users, and you don't
> need to cross JVM boundries, why use a DB at all? Why not just use a
> static
> concurrentHashMap and push the data back and forth that way? You'll have
> less overhead and one less configuration item to worry about.
>
> --- Pat
>
> > -----Original Message-----
> > From: djtrypt [mailto:[EMAIL PROTECTED]
> > Sent: Monday, November 21, 2005 11:33 AM
> > To: [email protected]
> > Subject: RE: Sharing an HTTPSession Between Apps
> >
> > well we looked into some In-memory DBs to store this and our current one
> > we are thinking of implementing is HSQLDB (hypersonic SQL DB) which is
> > 100% java based.
> >
> > Our app is still in beta phase so we wont have alot of concurrent users
> > for a few months till we release v1.0, so i think an in-memory would be
> > the easiest for our crew to set up, as we dont have a dedicated DBA at
> the
> > moment. SO if you know any let me know =P.
> >
> > Honestly after crunchin some number i dont either method we have talked
> > about would be a speed issue, as since the data is so nominal it wont
> get
> > stuck in bottlenecks.
> >
> > I think HSQLDB would be a fine match for us atm, as its very small
> (about
> > 115K) and each row costs 100bytes of memory. and we have no more then
> 12KB
> > of data to pass, so with that said even at 1000 concurrent users, thats
> > what 12 MB of RAM. I think we can afford that. =P
> >
> > however once we get an established Data Model going, ill prolly write a
> C
> > app to handle our Data Model more efficent then what will will start
> with.
> >
> > thanx for your time the stats are useful
> >
> > evan m rawson
> >
> >
> > -------------------- m2f --------------------
> >
> > Sent from www.TapestryForums.com <http://www.TapestryForums.com>
> >
> > Read this topic online here: <<topic_link>>
> >
> > http://www.tapestryforums.com/viewtopic.php?p=10986#10986
> >
> > -------------------- m2f --------------------
> >
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

Reply via email to