ASAP Performance & Responsiveness Is Poor (Doc ID 1990554.1)

Last updated on OCTOBER 21, 2024

Applies to:

Symptoms

ASAP performance & responsiveness is very poor.

Each Work Order has a single CSDL, and the CSDL has 2 ASDLs. We are injecting orders into OSM at a rate of 5 orders per second.

The parameters in ASAP.cfg are set to the values for a "Large" ASAP system, according to Chapter 4 of the Administrators Guide . The claim in that document is that "Large" system handles 20.95 orders/sec and 146.65 ASDL/sec but currently the average time spent in one of the ASDLs is 2.6 seconds and in another  is 47.81 seconds. Even more, the average time on WO/CSDL level is 153.81 seconds which is much more than the sum of the 2 ASDLs (2.6 s + 47.81 s = 50.41 s) that this Work Order consists of. The actual Java code in the processor (including the round trips to the network element) takes less than 1 second  and less than 4 seconds for another - this was confirmed by running a single order. However, when submitting multiple orders (1000), performance degraded significantly, both on ASDL level and on WorkOrder level. The network element SEACH_ADRN has 100 connections defined in the pool and our expectation is that the response time for 1000 orders should not be significant multiples of the response time for 1 order: seconds for 1000 vs 4-5 seconds for a single order. The hardware is very powerful and prstat shows CPU is at about 10-12% utilization total for all ASAP processes.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.