Home > Uncategorized > XIV Real Time Compression Q&A

XIV Real Time Compression Q&A

I had a customer recently upgrade their XIVs to the 11.6.1 code and wanted to know more about the real time compression that is now built into the code. IBM purchased the compression technology from a company named “Storwize” and adopted the name for their mid-range product family. One of the cool things that came out of that acquisition is the RACE engine that runs the compression algorithm.

Basically the compression is your standard LZ compression with some cool technology that keeps meta data about what you are writing. For XIV, this happens before the data is written to cache and turns all of those random writes into a sequential write. If you want to dig into the way RTC works check out this link.

The XIV team took the compression technology and integrated it into the XIV. The GUI even will show you an estimate on which volume will get better compression savings in order to fine tune your workloads. xiv_gui_rtc

Q. How does the compression work? Is it on all the volumes/pools?

A. Compression can turned on if a pool is set to “thin”. This thin pool then can have both thick and thin volumes in it. Once compression is turned on and licensed you can convert any thin volume to a compressed volume by right clicking on it and choosing compress. You can also tell XIV to compress all of your volumes or de-compress all of your volumes.

Q. Can I turn it off and on based on compression savings?

A. Yes, compression can be turned off on the whole box or just one volume. You get to decide based on the compression savings

Q. What size volumes can it handle?

A. The maximum is 10TB and the minimum is 52GB

Q. What kind of performance hit will it have on the XIV?

A. There are some benchmarks in the redbook with turning on compression that can help you decide on compression. Basically it comes down to two things: the workload and the model of XIV. If you look in the Compression Savings field and it is 30% or less then the you should not compress that data. If you have a new 314 model or 214 model compression can be turned on but 114 models need to be checked to make sure there is enough horsepower.

Q. What type of data does not compress well?

A. I get this question a lot. The basic answer is any time of data that has either already been compressed. Also backups seem to have lower compression savings. The better answer is always look at the compression savings in the gui and base conclusion on that output.

Q. Can I compress volumes that are mirrored?

A. Of course. Mirrored volumes can be compressed on both sides. If the mirror already exists then the mirror has to be broken, data compressed and the mirror copy has to be re-synced. We have seen a major performance improvement with compressed mirrored volumes as the amount of data being transferred is cut into half or less.

If you have questions about running compression on XIV or any other IBM platform leave a comment and I will try to answer them here.

Categories: Uncategorized
  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: