Highlighted
Absent Member.
Absent Member.
5750 views

Load Generator - RAM to User Ratio

Jump to solution
Good morning,

Does anyone have a document which spells out the appropriate RAM to User ratio for using a Load Gen for the various protocols?
0 Likes
1 Solution

Accepted Solutions
Highlighted
Absent Member.
Absent Member.

Re: Load Generator - RAM to User Ratio

Jump to solution
Hi,
I dont have a document regarding the same. But can you plz say me how many users are you trying to test and what kind of application you are trying to test.
Also plz mention that what is the protocol used for ur scripting. Probably there can be something, whcih i can help u

N rathi

View solution in original post

5 Replies
Highlighted
Absent Member.
Absent Member.

Re: Load Generator - RAM to User Ratio

Jump to solution
Hi,
I dont have a document regarding the same. But can you plz say me how many users are you trying to test and what kind of application you are trying to test.
Also plz mention that what is the protocol used for ur scripting. Probably there can be something, whcih i can help u

N rathi

View solution in original post

Highlighted
Absent Member.
Absent Member.

Re: Load Generator - RAM to User Ratio

Jump to solution
Memory calculation of the machine from where the Load generator generates the load:
Web protocol: 2MB memory required to execute 1 user
CITRIX protocol: 10 MB memory required to execute 1 user
AJAX protocol: 10 MB memory required to execute 1 user
SAP GUI protocol: 10 MB memory required to execute 1 user
Any other GUI protocol: 10 MB memory required to execute 1 user
Highlighted
Absent Member.
Absent Member.

Re: Load Generator - RAM to User Ratio

Jump to solution
Here is the offical HP stance on this:

Title: Is resource (cpu and memory) footprint information available to help in estimating the load generators required for load testing with LoadRunner?

Document ID: KM767319
Product: LoadRunner
Version:

Updated: 2009-Sep-21

Summary:
Accurate resource footprint information is not generally available. It is recommended that existing resource usage data is used and/or specific load tests are conducted (e.g. single Vuser) to allow resource usage to be measured.

When planning a load test it is necessary to consider what resources (cpu and memory) will be needed to generate the required load.

This information can then be used to choose suitable load generator systems and determine how many of these systems will be required.

Is resource footprint data available for the various protocols that are supported by LoadRunner to allow load generators to be sized accordingly?

Solution:

Given the range of protocols that are supported by the LoadRunner product and the variety and complexity of scripts that may be used in load testing, it is not possible to provide accurate resource footprint information.

Instead it is recommended that when resource estimates are required existing resource measurements are used (from previous load tests) and/or specific tests (e.g. single Vuser) are conducted to enable the measurement of the resources being used.

A specific test approach is recommended as resource footprints are highly dependent on the specific application and systems in use. Such an approach will help ensure that the specific aspects of the IT infrastructure to be load tested are taken into account.

It is also recommended that resource usage is monitored during load testing to verify that the estimates match with actual experience. It is important to ensure that load generators do not become overloaded as this can render a load test invalid. For instance plan and monitor cpu usage to ensure that it does not exceed 80-90% of that available. If a load generator is allowed to become cpu bound then this may affect, for instance, load test response times.

If the protocol under test is threadsafe then run each Vuser as a thread to allows resources to be conserved while supporting more Vusers.

My thoughts:

We are testing the FTP protocol with one controller and three Load Generators of 250 Users each.

My take from what HP explained is that each script will determine the amount of VUsers which a Load Gen can handle based on the ease or complexity of the script.

So the trick would be to test the Load Gen with one user to determine the CPU utilization of the LG then move from their so as not to exceed 80 to 90% of CPU.

Your thoughts please?

0 Likes
Highlighted
Trusted Contributor.
Trusted Contributor.

Re: Load Generator - RAM to User Ratio

Jump to solution

Is there any matrix availbel to get the memory/cpu foot print for each protocol level

Eg: Web http html  - 5MB for one user  etc...

0 Likes
Highlighted
Acclaimed Contributor.
Acclaimed Contributor.

Re: Load Generator - RAM to User Ratio

Jump to solution

Hi @saravanansriniv,

We do not publish these spreadsheets anymore since they rely on too many factors (script logic, application under test, LG sizing) and are not useful as a general sizing reference. They are good for our internal testing to ensure footprint does not rise between one version to another. In the past when we shared it, customers escalated about inaccuracies, hence we stopped publishing it.

Regards,
Shlomi

0 Likes
The opinions expressed above are the personal opinions of the authors, not of Micro Focus. By using this site, you accept the Terms of Use and Rules of Participation. Certain versions of content ("Material") accessible here may contain branding from Hewlett-Packard Company (now HP Inc.) and Hewlett Packard Enterprise Company. As of September 1, 2017, the Material is now offered by Micro Focus, a separately owned and operated company. Any reference to the HP and Hewlett Packard Enterprise/HPE marks is historical in nature, and the HP and Hewlett Packard Enterprise/HPE marks are the property of their respective owners.