ESX 4.1 Memory Compression

Hi there, probably most of you are just thinking about the holidays, right ?
Well, since I’m still working, I’m also thinking about the holidays, but also on a new post.

This time I will write about VMWARE memory compression…it’s a new feature on ESX 4.1

We can think of it as a new layer of memory management.
At memory overcommit technology hierarchy it’s now placed between balloning and swapping.

1 – Trasnparent page sharing
2 – Balloning
3 – Memory Compression
4 – Disk swapping

So what happens is when ballooning is not capable of managing the workload instead of swapping the memory to disk ESX attempts first to compress the guest’s memory.
So the pages can be compressed and stored at the main memory.
The next access to that memory would require a memory decompression.

But why is that better/faster ?
Because the compressed memory still in main memory and a compress/decompress process is a lot faster than swapping in disk.

See the performance bellow:



Also If you want to read more Memory Resource Management in ESX 4.1 , check this out.

Happy New Year !!!!