VISIBILITY BLOG

header-picture

As applications are deployed to the cloud or even across complex intranets, the task of performance tuning often seems like more and more of an art form.  But if the process of tuning is done methodically, you can treated more as science and find and solve most any performance problem that can occur.

First, identify some benchmark problematic transactions. These are repeatable transactions that can be timed with a stopwatch and are performing consistently slower than they should. Then compare these timings to what is expected of the application in an ideal setup.  Software companies like ours, have ideal set ups which mimic customers in terms of size of data and load that are optimized. Your timings should be as good as the software company’s ideal setup.

For example, if you pull up a transaction maintenance form and it takes > 5 seconds consistently, while it takes < 1 second to do pull up the same transaction screen on the ideally optimized configuration, you have a problem and have identified a benchmark which you can then use to troubleshoot and confirm when the problem has been resolved. When it runs < 1 second, your problem has been solved.  Often a performance correction will impact lots of transactions in the system, not just the identified benchmarks. 

Next, you need to begin the troubleshooting process. The most methodical approach is to systematically narrow down and eliminate all of the high level components. Components include the database server, the web application server, the client, the browser settings, the internal network, and the external network if Internet is used.  This process of elimination can be as simple as using the browser from the application server to eliminate the client PC, or as complex as changing your virtual database server to ride on the same physical box as your application server to reduce and eliminate network issues as a possible cause.

Once elimination has occurred, you can examine more closely the remaining components for problems/possible bottlenecks. Checklists should be developed (e.g. check memory and CPU usage, check Disk space considerations, check anti-virus software, trusted sites and browser settings, database settings on memory usage and disk, temporary storage, network speed, internet connection speed, etc).  It only takes one bottleneck to cause performance issues.

By creating a methodical, scientific approach to troubleshooting performance in today’s complex application world, you can make tuning more of a science and less of an art, and have a much better success rate at identifying and solving any performance issue that may arise.

To read more about Visibility technical support and services, click here

Topics: Services


Recent Posts

Streamlining AP Operations: The Power of the Tipalti Partnership

read more

ERP Best Practices: ERP Data Cleaning

read more

The Ultimate ERP Wishlist

read more