For example: An applications subscriber base has tripled since it
was deployed causing increased message volume resulting in a queue
backlog and increased delivery latency. Multi-threading the queue’s
consumer easily removes the blockage and restores bandwidth but, as
a result the increased processing capacity has now overwhelmed the
downstream recipient consumer service.

Truly simulating production load can sometimes be a challenge for
QA organizations. However, application instrumentation, monitoring
and tunable settings (our ‘trick’) has helped us ensure that we can
detect and address live production issues before they cannonball
profitability with embarrassing SLA violations and costly charges.

In our case value (message latency) was definitely linked to service
quality (application stability). It has been vital to be able track,
analyze and thwart potential issues by having many dynamic service
configuration options (e.g. a max thread count setting) to ensure that
our customer’s varying needs are always met. When we have satisfied
larger population of our customer’s needs the overall perception of
service’s value and its profitability has also increased.