Linux Application Memory usage

Shell Programming and Scripting
1

Hello

We are in the process of migrating few of our applications in our linux boxes to new linux box to streamline our applications . In this context , i would like to know how we can calculate a particular application is used .?

This data will then be used to select which applications need to be moved to the new box . Could some one please let me know how we can get this information ?

Regards
Ron T

2

Hi Ron T.,
I have no idea how migrating an application is going to streamline that application. Can you shed some light on how this might happen?

With this in mind.. How would one calculate a particular application (as you requested):

You must know a lot more about your applications than the nothing you have told us. Based on what you have told us, I have no idea how we could help you calculate anything??? Nor, do I have any idea what you might be trying to calculate???

Can you provide any more details that might help us help you?

3

Hi Don-

In the existing boxes , we are running short of memory and the we have same applications running for multiple countries . we decided ,to remove few applications from the existing boxes and going forward the boxes will be country specfic i.e one box will have only one country applications .

Thanks
Ron T

4

This is all very interesting. But, I don't see how it relates to the questions raised in this thread.

What does:

mean?

If you are creating one box for each country to be processed, doesn't that mean that each application will be running on each box? If you don't have enough memory to run all of your applications on one box; how does creating multiple boxes each of which is running all of those applications reduce the amount of memory you need?

5

In future one box will have one country applications & all applications of that country is not going to run in one box. We now need to calculate memory usage of each of the applications & for this what would be the best way to go forward ?

6

Let me see if my understanding comes close. On one single server, you have several different (database?) applications, each of which is running on different datasets for different countries; it now becomes overloaded and causes problems, probably performance wise. Can you quantify the problem?
So you have APP_A running on DB1, APP_A running on DB2, APP_A running on DB3, then APP_B running on DB4 (or is it DB1 again?), APP_B running on DB5 / DB2, etc.
You might use tools like vmstat , ps , or top to extract memory consumption for processes, and there might be others as well. Use with care, and take a few snapshots over time to account for load variations.
The difficulty of answering your question lies in e.g. shared memory, as multiple application instances might work from one single code copy in memory, and database systems tend to utilise shared memory as well for effectiveness.
I'm afraid there's no easy, general way to get at your desired numbers; you might want to hire an expert for this analysis.

7

Ok. It depends US & Canada share the same Db where as for e.g Mexico has a different DB . I am currently using the below command to see the memory usage

ps aux --sort rss

so does this give a approximate memory usage , the box size is 32 GB . Here what should i choose VSZ or RSS ?

Thanks
Ron T

8

Take RSS.
VSZ is too unprecise, especially with Java applications.