Additional Blogs by Members
cancel
Showing results for 
Search instead for 
Did you mean: 
Former Member
0 Kudos

With the rising popularity of virtualization and cloud computing, we are once again faced with some very basic challenges when it comes to performance and scalability.  It's the same challenges we've always had with "bare metal", but made more complex by virtualization and the sharing of physical resources. This series of blog posts will explore ways to compare "apples to apples" when it comes to measuring the performance of your SAP environment on "bare metal", in a virtual machine or on the cloud. 

To accomplish this in a way that is easily repeatable and which anybody can use, we will use a series of steps that any SAP application server, including the trial version, can execute.  The number of times one virtual user can complete this sequence in a minute will be basis for comparing performance in different environments.  What is so important about that number is how many things it takes into account or factors in, from the OS to the database to the network stack.  Repeating the same test multiple times in a row should yield the same number +- 1% or so.  Why is this?  Because of other things that may be happening on the server, which is the reason to make sure the computer is idle when the test is run. 

Increasing the number of virtual users will produce a proportional, linear curve until either a CPU or bandwidth limit is reached, at which point the "hockey curve" of performance degradation appears. 

Running the same test on bare metal, in a virtual environment and on the cloud will show the differences in performance and help us to manage something that was previously difficult to impossible.  But these kinds of standard performance rating tests have to become common if we are truly to exploit the vast potential of virtualization and cloud computing!

In the next entry I will post the results of the first test and show the steps I am using to create it.  It would be fun if other people would run the same test on their system and post the results - over time, we could start seeing some patterns that would otherwise totally escape us!