-
- Enthusiast
- Posts: 91
- Liked: 5 times
- Joined: Aug 26, 2014 1:33 pm
- Full Name: Dave Twilley
- Contact:
Azure Blob Single Instance / Deduplication?
If I use Azure Blob as a backup copy repository, will the data be single instanced?
eg. We have 10TB of VMs and we want to have a week end copy retained for 4 weeks, would this consume 10TB in Azure Blob or 40TB in Azure Blob?
ie. will it reference the blocks already uploaded if they've not changed.
We accept there will be some additional storage cost based on rate of change within this data.
Without this, you can see why we're worried about 10TB with 4 weekly, 12 monthly and 7 yearly backups as it will grow massively out of control without single instancing / deduplication of blocks.
Many thanks.
eg. We have 10TB of VMs and we want to have a week end copy retained for 4 weeks, would this consume 10TB in Azure Blob or 40TB in Azure Blob?
ie. will it reference the blocks already uploaded if they've not changed.
We accept there will be some additional storage cost based on rate of change within this data.
Without this, you can see why we're worried about 10TB with 4 weekly, 12 monthly and 7 yearly backups as it will grow massively out of control without single instancing / deduplication of blocks.
Many thanks.
-
- Chief Product Officer
- Posts: 31645
- Liked: 7139 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: Azure Blob Single Instance / Deduplication?
Yes, the data is single instanced aka forever-incremental approach: only changes since last backup are captured, transferred to and stored in object storage. Weekly, monthly and yearly backups are therefore "spaceless" i.e. they don't consume any storage space. They are merely pointers to existing objects brought over by previous backup runs.
The actual consumption will depend solely on your data change rate: whether those VMs are running workloads that are relatively static, or constantly produce large amounts of unique data. Across all of our customers, average daily change rate is in the area of 3-5%. So you can use this for initial disk space requirements estimation. Remember to divide the resulting number by 2, which is an average backup compression ratio.
The actual consumption will depend solely on your data change rate: whether those VMs are running workloads that are relatively static, or constantly produce large amounts of unique data. Across all of our customers, average daily change rate is in the area of 3-5%. So you can use this for initial disk space requirements estimation. Remember to divide the resulting number by 2, which is an average backup compression ratio.
-
- Enthusiast
- Posts: 91
- Liked: 5 times
- Joined: Aug 26, 2014 1:33 pm
- Full Name: Dave Twilley
- Contact:
Re: Azure Blob Single Instance / Deduplication?
Perfect thanks, Veeam FTW
-
- Enthusiast
- Posts: 91
- Liked: 5 times
- Joined: Aug 26, 2014 1:33 pm
- Full Name: Dave Twilley
- Contact:
Re: Azure Blob Single Instance / Deduplication?
Do you have any guidance on rate of change over a longer period of time. Obviously using 3-5% * 31 days isn't realistic for a monthly rate of change as its likely to be the same data which is changing in most businesses.
With 3-5% rate of change we can calculate x number of days retention but nothing past that. I guess we'd need an average monthly and yearly rate of change too.
With 3-5% rate of change we can calculate x number of days retention but nothing past that. I guess we'd need an average monthly and yearly rate of change too.
-
- Chief Product Officer
- Posts: 31645
- Liked: 7139 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: Azure Blob Single Instance / Deduplication?
It's utterly unpredictable due to being extremely workload-specific. But I would advise against planning for less than 3% per day across 31 days in any case as an average across the entire environment.
Also, if you keep backups for many months/years then you should definitely consider using Archive Tier as well. This makes a huge difference as it comes to the long-term storage costs.
Also, if you keep backups for many months/years then you should definitely consider using Archive Tier as well. This makes a huge difference as it comes to the long-term storage costs.
-
- Enthusiast
- Posts: 91
- Liked: 5 times
- Joined: Aug 26, 2014 1:33 pm
- Full Name: Dave Twilley
- Contact:
Re: Azure Blob Single Instance / Deduplication?
When you say:
I appreciate there re no guarantees, its tied to the type of workload, i'm just looking for some ranges on which to base some estimates, i'll always tell a customer it could be significantly more or less than the estimate.
Are you therefore warning not to use less than 3% x 31, ie 93% per month!? That would suggest you're saying that we should expect an entire environment to change in a month.But I would advise against planning for less than 3% per day across 31 days in any case as an average across the entire environment.
I appreciate there re no guarantees, its tied to the type of workload, i'm just looking for some ranges on which to base some estimates, i'll always tell a customer it could be significantly more or less than the estimate.
-
- Chief Product Officer
- Posts: 31645
- Liked: 7139 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: Azure Blob Single Instance / Deduplication?
That is correct. Unless your customer has a quiet environment with fairly static workloads, I would not plan for less. Remember this is average across the entire environment. Active database servers will have almost the entire disk content changed every single day! And you will need a lot of "very quiet" VMs to balance that out and get to 3% daily change rate across the entire environment.
Who is online
Users browsing this forum: No registered users and 14 guests