Comprehensive data protection for all workloads
Post Reply
kapsula
Lurker
Posts: 2
Liked: never
Joined: Oct 09, 2018 7:58 am
Contact:

Deduplication Best Practices on Windows 2016 scale-out repo

Post by kapsula »

Hi everyone,
a brief overview about infrastructure
- around 1600 VM (230TB) on vSphere
- 8 phisycal proxies (4 for each site)
- Primary backup model: forever incremental w/ active full on 140TB scale-out repo (actually simple NTFS, we're planning to move to a REFS for resiliency and fast-clone)
- Secondary backup: backup copy with archival option enabled (4 weekly fulls, 12 monthly fulls, 2 yearly fulls) on 100TB scale-out DEDUPED repo (NTFS on Windows 2016)

First of all I'll be glad to receive your feedback about such approach.

Moreover I've a couple of doubts about deduped scale-out repository.
Following the best practices guidelines (https://bp.veeam.expert/architecture-ov ... pe_windows and https://www.veeam.com/blog/data-dedupli ... veeam.html) I'm not sure about following settings

1) "Enable inline data deduplication" and "Compression Level" settings: bp.veeam.expert set to disable the first one and optimal the second one; Andrew Zhelezko's blog post instead enable inline deduplication and set compression to "none". What's the right way?

2) on bp.veeam.expert, talking about Align backup file data blocks, the bp say "Enabled - only if repository uses fixed block size deduplication (almost never true)". What does it mean?

3) about garbage collector schedule, bp say to change it from weekly to daily, but I found an MS kb warning about full GC can impact I/O performance and incremental backups size (https://support.microsoft.com/en-us/hel ... cause-perf and https://social.technet.microsoft.com/wi ... rview.aspx).
By design every 4th Regular GC job the system runs a FULL GC job: it would be better to completely disable Full GC by adding the following reg key?

Code: Select all

reg add HKLM\System\CurrentControlSet\Services\ddpsvc\Settings /v DeepGCInterval /t REG_DWORD /d 0xffffffff
Thanks for your attention!
Rick.Vanover
Veeam Software
Posts: 712
Liked: 168 times
Joined: Nov 30, 2010 3:19 pm
Full Name: Rick Vanover
Location: Columbus, Ohio USA
Contact:

Re: Deduplication Best Practices on Windows 2016 scale-out repo

Post by Rick.Vanover »

(redacted)
Gostev
Chief Product Officer
Posts: 31803
Liked: 7298 times
Joined: Jan 01, 2006 1:01 am
Location: Baar, Switzerland
Contact:

Re: Deduplication Best Practices on Windows 2016 scale-out repo

Post by Gostev »

1. Most important is to set compression to "none". Disabling inline dedupe should may make the processing slightly faster, but your mileage may vary.
2. Disable this, since Windows Server deduplication uses variable block size dedupe.
3. I would trust BP on this one.
kapsula
Lurker
Posts: 2
Liked: never
Joined: Oct 09, 2018 7:58 am
Contact:

Re: Deduplication Best Practices on Windows 2016 scale-out repo

Post by kapsula »

Thank you Gostev!
Post Reply

Who is online

Users browsing this forum: Google [Bot] and 113 guests