We had to deal with that situation alot when I first came to work at MD Anderson Cancer Center a long time ago.
We solved most of our resource starvation issues by breaking interfaces into smaller separate sites and processes based on resource demands until we achieved the parallel throughput that was acceptable.
Some of our sites don’t even have any inbound or outbound interfaces to any foriegn systems and are only used as distrubution sites for internal purposes.
We have been able to extend the life of our server by doing this and surpassed any throughput threshold I would of thought possible; today we move about 13 million messages daily with horsepower to spare and no more back logs caused by resource contention.
The methods we used for breaking things up is what we call:
– a TCP/IP hop for hoping a message from one process to another in the same site.
– a TCP/IP jump for jumping a message from one site to another on the same server
– a TCP/IP leap for leaping a message from one site to another site on another server
Another sitiuation that ate my lunch was an interface that was using something like disposition OVER that caused a perpetual loop where one message was getting sent 80,000 times before it got unbearable and had to be remediated.
In other words look for dangerous/risky dispositions which the clovertech forum can comment on what those might be because it might be PROTO and not OVER I just don’t remeber but there are some dispositions that can create nightmares if not used appropriately.
Russ Ross
RussRoss318@gmail.com
.
