My understanding of these parameters and others has been the result of an extensive background in UniData and UniVerse tuning. Over the years I have spoken to many, if not most Rocket Software engineers as the result of dealing with various client issues. This is one of those things I picked up over time and with experience.
Our software, Mercury Flash (www.mercury-flash.com) has an AutoTune feature that collects various details about a system and then uses formulas we have created to tune some of the parameters found in UVCONFIG and UDTCONFIG files. These specific parameters are a few that we have recently started designing a formula for to determine the optimal tuning values.
Due to the differences in systems such as Memory, CPU performance, type of storage, SSD, vs RAM disk, SAN and NAS storage and conventional hard drives with differing spindle speeds in addition to various O/S configurations, identifying the proper tuning parameters for any system can be challenging. This process would be a lot easier if we had access to the UD and/or UV source code, but we don't so the best we can do is collect information from Rocket Support and continue to solve various client issue and collect information to arrive at a solution that fits the majority of cases. It is also important to remember, that what works on one system, may not work on another. It is these variations in performance, that provide clues as to how the various parameters work.
Most end users are dealing with a very narrow focus when it comes to performance tuning. They generally only see their system and work within those limits. At Paradigm Systems we have been building a knowledge base dedicated to performance tuning and are constantly working to tune our formulas. Performance tuning is not a one and done endeavor, over time user load, system age, database size and software changes (both user software and Database software) affects performance, so performance tuning is a continuous process therefore tuning parameters change over time.
As for parameters like SCRSIZE, simply changing a parameter value may not be instantly evident. Some changes are only effective in specific situations, other times changing a parameter can be instantly noticeable such is the case with SELBUF. The key is to monitor the entire system, looking at various metrics over time as some changes are cumulative and are only visible with multiple users under heavier load.
I hope this answers your question and I know for some it will only create more questions.
------------------------------
Jay LaBonte
President & CEO
Paradigm Systems, LLC
[Phone] |
https://www.paradigm-systems.us/------------------------------
