No subject

anymore ...

;*)

... but seriously ...

I do sometimes wonder ... is it "bad" to solve these problems in house ...
because I rarely see questions of this nature illicit a "hey ... we wrote
that ourselves and it wasn't that tough ... here are some things to watch
out for ... and good luck."

Good Luck,
Tony

PS. Here is an 'off the top of my head' list of things to watch out for ...

1. you need to make sure that the service you are handing out a proxy for is
actually running when you hand it out to the requester. (think 'ping')

2. The "services broker" server that is responsible for managing the proxies
to the services you are trying to load balance will go down. Have a plan. I
handled this by having the services themselves actually ping the directory
server periodically ... and if they find out it has gone down ... they will
re-register themselves. This is all encapsulated such that the thing that IS
the service has no knowledge of any of this. I thought about actually
building a 'failover' services broker ... but the services broker itself is
pretty simple ... and we will be running it on kick butt, redundant hardware
... so I figured if we run into a problem where we can't get it back up and
running pretty quickly ... then we probably have a serious problem that even
a 'backup' broker would not solve.

3. All things in our world that want to be 'services' such that they can be
load balanced ... have to satisfy a BASIC contract to their services
container that manages them such that they provide a basic metric that the
container can query  which indicates "how busy they are". The container than
advertises this to the services broker ... which uses this metric to balance
the load across all replicates.

---
To unsubscribe, mail advanced-java-unsubscribe at xcf.berkeley.edu
To get help, mail advanced-java-help at xcf.berkeley.edu