We're building a "business-critical" web app for a small business. They
will rely heavily on this app every day (all day), so obviously down-time
(of 2 hours+) is not an option. The only way to solve the problem is with
clustering. However, cost is also an issue - as they're a small business.
(Don't worry, I'm getting to the question) So using three or more
servers is just too heavy a cost for them (well, three might be possible,
but 4+ is not). It seems to me that there might be a chance to use Jini in
this situation - given the broadcasting capabilities and leasing.
Is something along the following lines possible?
HTTP Client
|
router
|
-----------------------
| |
Server 1 Server 2
Where Server 1 and 2 both have Jini Discovery services running, and they
register with each other for sharing/replication of data.
My questions:
1. Would this need a dispatcher/server right behind the router or is it
possible to have the router send data on to both of the servers (even if it
alternates randomly, but knows they're up or down)?
2. Would this type of fail-over/load balancing be possible with Jini and
would it even be of benefit?
The main problem we're trying to solve is making fail-over possible for a
small company - i.e. without the high cost of hardware/hosting services. Is
this possible with jini? Am I looking at this all wrong?
Thank you for any help you can give.
|
Similar Threads
Linear Mode
