Prophet 21 Performance – Desktop vs Web Client

The more I talk to users and system administrators about the Epicor Prophet 21 web client, the more I get asked about Prophet 21 performance in the web client versus the traditional desktop client.   Today, we are going to do a side by side comparison of Prophet 21 performance using loading times for common windows in the web client versus the desktop client.

The Prophet 21 Performance Test Environment

For this Prophet 21 performance test, we are using the live environment of a 200 user P21 deployment.  The desktop client is a published app via Microsoft Remote Desktop Services.  The web client ran on Google Chrome.  Both the middleware server and the remote desktop server have the exact same hardware specifications.

  • Host CPU: Intel Xeon E5-2667v4 (8 core)
  • RAM: 20Gb
  • Storage: SSD drives in RAID 10
  • Prophet 21 version: 2017.1.2950
  • Operating System: Windows 2008 R2
  • SQL Version: 2016 with CU5
  • Hypervisor: vmWare ESXI

We ran web client via the SSL port.  Also, we did not load P21 “over the Internet”.  We used the web client only within the LAN environment.  Finally, anytime a window was loaded in the web client, it was opened in the current tab.

For timing, I used a simple stopwatch app in Windows and situated the start button very close to the P21 menu.  I believe there was a fairly consistent 1/4 second delay for both the web client and the desktop client. We conducted each test 3 times, and the results each time were very similar.

Opening Order Entry

Like it or not, the speed at which the Order Entry window loads is the de facto measurement of Prophet 21 performance in the eyes of many users and administrators.  You are unlikely to get into a conversation about Prophet 21 performance without talking about how fast OE opens.

For this test, we disabled all OE panels.  There are significant DynaChange Screen Designer modifications to the UI.  Additionally, we enabled DynaChange caching.

The OE window opened in 8.6 seconds on the desktop client.  It opened in 11.2 seconds on the web client.

Loading an Order from an Open OE Window

For this part of the Prophet 21 performance test, we loaded a 35 line sales order. It is filled mostly with items that are assemblies.  In each case, the Order Entry window was already open.  No existing order was previously loaded.

The results of this test were that the order fully loaded in 3.1 seconds on the desktop client.   The web client loaded the same order in 6.9 seconds.

Opening Item Master Inquiry

For this test, we opened Item Master Inquiry with no other windows already open.  In both cases, we opened the screen by clicking from the main menu of Prophet 21.

Prophet 21 performance test result:  Desktop client loaded the window in 3.6 seconds.  Web client loaded the window in 3.7 seconds.

Opening Customer Master Inquiry

In this final test, the Customer Master Inquiry window opened with no other windows open.

The desktop client loaded Customer Master Inquiry in 3.2 seconds.  The web client loaded the same window in 3.6 seconds

Last Thoughts on Prophet 21 Performance Testing

This Prophet 21 performance test was neither highly scientific nor perfectly controlled.  That said, it is representative of how the average user may perceive the system.  A true assessment of Prophet 21 performance requires more extensive testing. This is particularly true in the order entry screen, .  Perhaps in a future performance test, we can also look at OE load times under varying conditions. This may help determine the effects of screen changes, caching, business rules, panels, and other optional features on the overall system performance.

Finally, as you are planning your Prophet 21 web client deployment, it is a good idea to do some performance testing of your own.  The more ways you find to squeeze performance out of the system, the happier your users will be.

For more articles, visit our Prophet 21 Knowledge Base.

Posted in Prophet 21, Web Client Stuff and tagged , , , .