Perf Data on LSM in v5.3

[Casey Schaufler]

On 1/14/2020 12:15 PM, Wenhui Zhang wrote:
> Hi, Casey:
>
> I just performed a performance check on 
> 1. v5.3 with DAC only, and 
> 2. v5.3 with DAC and MAC framework, an empty-policy enabled in sub-modules(e.g. selinux)
> It seems like the downgrade is quite obvious. (here the perf is described in range 0 - 1, 0 is 0% and 1 is 100%).
> I am using a typical laptop for test,  4 2750MHz CPU, with SSD and 8G memory.
> Hope this might raise the performance issue.

Thank you for doing this. Unfortunately, your graph is unreadable.
Could you provide some more detail about your results?

>
> The last paper mentioning about performance of LSM was about 20 years ago, where Chris was using a setup of 4 700MHz CPU, 128MB memory with Disk. 
>
> Due to the change of hardware, performance on filesystem changes a lot.
>
>
> sec1.png
>
> -- 
> V/R,
>  
> Wenhui Zhang
>  
> Email: wenhui at gwmail.gwu.edu <mailto:wenhui at gwmail.gwu.edu>
>            Telephone: 1-(703) 424 3193
>           
>  
>  
>  
>
>