Solved

Cloud Connect

  • 2 May 2019
  • 1 reply
  • 7124 views

Badge +1
Hello - I'm new to Nutanix and am in the middle of taking the online 5.5 course in the education portal. I have a simple question about Cloud connect:

In the section about Cloud Connect features the very first thing you read is "data transmitted is already deduplicated and users can choose to enable compression on the local storage container.."

Yet when I look at the section regarding General Recommendations and Limitations I see the following: "It is not recommended to enable deduplication on the source cluster..."

Can you please help me understand this? This looks like a complete contradiction.
icon

Best answer by ddubuque 2 May 2019, 20:31

Sometimes with documentation you'll find remnants that are outdated.

Unless an application such as Microsoft Exchange explicitly call out for storage to not have dedupe turned or or compression, it's always a best practice.

I would raise this up to support for a clear and official stance, but my personal opinion would be to turn on dedupe and compression. The caveat is understanding that you will need to increase RAM on your CVMs of you enable it. Compression on a highly active container can incur higher CPU overhead, so setting post-process compression is recommended. I have some clusters where we set it to compress almost a day after the initial write so I don't have the CPU overhead as high on my CVMs.

Hope that helps.
View original

1 reply

Userlevel 3
Badge +10
Sometimes with documentation you'll find remnants that are outdated.

Unless an application such as Microsoft Exchange explicitly call out for storage to not have dedupe turned or or compression, it's always a best practice.

I would raise this up to support for a clear and official stance, but my personal opinion would be to turn on dedupe and compression. The caveat is understanding that you will need to increase RAM on your CVMs of you enable it. Compression on a highly active container can incur higher CPU overhead, so setting post-process compression is recommended. I have some clusters where we set it to compress almost a day after the initial write so I don't have the CPU overhead as high on my CVMs.

Hope that helps.

Reply