7 Posts
1
10604
Creating a Repurpose SQL snapshot on XtremIO - all databases
Hey guys,
I'm a Database Developer trying to get a handle on Appsync so we can start using it with our XtremIO SAN, and take away some of the hassle from our overworked Sysadmin, who has been mounting Protect (Service Plan) copies, unmounting and complaining of having to manually unmapping and expiring the copies.
I'm trying to perform a regular repurpose copy for development, but I am finding when I perform a snapshot repurpose, that it says in the details:
- Backup if database \ completed successfully
- VSS application thaw succeeded on host .
- Created snapshot-set SnapshotSet.1543461779789 using the CG _ All_Databases on XtremIO CKM00163500678.
- Created XtremIO snapshot ..snap.20181129_144135.g1 of XtremeIO volume on XtremIO CKM00163500678.
The problem here is, although I am on the "User Databases" page of "Copy Management", I am selecting a single database, and I want to just repurpose that single database - and instead, this makes a g1 copy of all the databases on the server.
Is there something I'm doing wrong here - why is it choosing All Databases when I am after the one?
We're also finding that in the XtremIO utility, it is reporting the copy is the full size of the original - so, for a 50gb database, it'll report that an additional 50gb is used for each copy we make - eg 3 copies is 150gb on our XtremIO. My understanding was that it was a point-in-time diffed copy and it would only accrue additional space as the two copies became different from each other?
schifa
82 Posts
1
December 4th, 2018 06:00
There is no right or wrong way to accomplish what you are doing. The two types of workflows can essentially support both use cases, i.e. protection and repurposing.
The protection seems straightforward for you. The service plan can be utilized to keep any number of copies, ready to be utilized for restoring. The time to restore I cannot state, but from working with many other customers, AppSync has reduced those times significantly..all dependent upon the environment of course.
For the other use case, repurposing, you can utilize either type of workflow. I suggested service plans, due to the way your applications are laid out. If there are multiple applications within one CG, then you really do not have a choice, unless you can break them up at the storage layer. If you can create unique CGs per application/database, then the repurposing workflows are certainly more conducive for your purposes.
Utilizing the repurposing workflows will provide the ability to create 1st and 2nd gen copies. The 1st gen could also be utlized for a restore, so long as it is never mounted, and the 2nd gen copies happen near instantaneously, though the mount operations may take longer. The 2nd gen copies can be divvied out to the developers, who all will have a working copy of the 1st gen. This solution will only affect the application once. This is of course, as we have discussed, dependent upon the layout of the CGs, as the repurposing workflows perform copy management differently than the service plans.
AndrewTobin
7 Posts
0
November 28th, 2018 21:00
Hey folks,
A lot more researching and working out my terms and how both XtremIO and AppSync work (and a lot of EMC whitepapers) later... and I believe I'm asking -
When I perform a regular "Protect" copy of a database, I get a Volume snapshot of that database.
When I perform a repurpose copy of a database, I'm getting a Consistency Group copy
Is there any way to ask the repurpose workflow to only back up the volume with the database I'm after, as opposed to a full CG?
Why would the selection of what is being snapshot differ between making a Protect copy against a Service Plan, and a Repurpose Workflow?
schifa
82 Posts
0
November 29th, 2018 05:00
This is how AppSync behaves and there is no way to change it. For the protection workflows, i.e. Service Plans, AppSync can subscribe the User database folder, which will protect all databases on that particular SQL instance. Alternatively, you can subscribe only one database, which will then just protect that one database.
For the repurposing workflows, you cannot repurpose more than one database at a time. The intention of that workflow is to create a copy of a single database, and repurpose it for other purposes. The repurposing workflows do utilize CGs for efficiency purposes, rather than individual LUNs, as the protection workflows support.
There should be no reason AppSync creates an entire copy of a LUN. AppSync uses XtremIO APIs which are not thick and will take snapshots of the LUNs. XtremIO does not provide an option to create anything other than snapshots, so I recommend reaching out to your account team, and asking them to consult with an XtremIO SME.
AndrewTobin
7 Posts
1
November 29th, 2018 06:00
Thanks for the response.
We are working with consultants- but it's been a difficult process and I think some of that is my lack of understanding and finding a common way of communicating what we want.
The service plans, I think I have a pretty good handle on, and what you're saying makes sense to me.
The repurposing - even though you can only repurpose one db, the act of snapshotting the CG makes a copy of all the server's databases, because they're all (the LUNs?) in the same Consistency Group. It still only creates a repurposed copy in AppSync for the database we're repurposing, but there are g1 copies for the other databases on the XtremIO tool.
Would it make more sense to have each database/mount point (LUN?) on its own Consistency Group instead of one group for the whole server?
Sent from my iPad
schifa
82 Posts
0
November 30th, 2018 05:00
I have two questions...
Most often, both workflows can help you achieve what you want, so the first question will help me understand if you can achieve what you need from the protection workflow, i.e. service plans. If you must use the repurposing workflow, and there is no business justification for utilizing one CG for all DGs, then you can break them up. This would simply be more work for you, so first, we need to understand your copy management needs.
AndrewTobin
7 Posts
0
December 2nd, 2018 21:00
Sorry if I'm being unclear.
1. We're currently using the service plan/protection workflow and mounting that as our development and test environment. It was my understanding from reading the documentation that, for this purpose, we should be using the repurposing workflow.
2. I'm not sure we do - this was how it was set up a long time ago, by the previous administrators - It is my thought that if it was the wrong workflow, or not set up appropriately, that it was part of why we have had issues with it.
Basically, all I am looking at is:
AndrewTobin
7 Posts
0
December 4th, 2018 14:00
Thanks schifa , that makes sense. Since a volume can exist in multiple CGs, I may test seeing if it still repurposes all of the "all databases" group if I repurpose - I'm assuming it will, and we may need to break those up.
I appreciate all the help and advice
norman.j
2 Posts
0
January 12th, 2019 23:00
How a “Restore” is done in XtremIO?
karlalujan
1 Message
1
January 21st, 2021 02:00
I have the same question " How a “Restore” is done in XtremIO? "
davidwar1
1 Message
0
August 16th, 2021 09:00
What's causing you to rebuild replicas? Depending on your environment and workflows, you should be able to maintain individual replicas that can be synced with the parent database.