Async replication and SQL Server consistency

[andrea.caldarone]

Hi all,

if I had StarWind Virtual SAN in both my production site and in disaster recovery site, and a lot of SQL Server on the top of it. I know that SW VirtualSan does block level replication (deduplicated, compressed etc.) but I have to know if it guarantes consistency in my database, let's make an example:
I have 2 LUN, one for data files and the other for log files, serious storage replication technology (for example EMC Recover Point) make "consistency group" where the order of writes in group of LUNs are guaranted to be replicated in the same order in the DR site. Has StarWind similar function?

[Anatoly (staff)]

Hello Andrea,

Thanks you for sharing your question with StarWind community.

You have a guaranteed replication to the DR site, and the data blocks will be written to DR site in the same order so as they appear on the primary location. The trick here is that replication runs on the background, thus it doesn`t affect the primary storage tasks processing speed. Also If the data on the primary site gets corrupted for any reason (malware, human factor, etc.), then it will StarWind allows to recover to any point in time from snapshots located on DR, unlike traditional DR solutions. Obviously StarWind guarantees that data will be delivered and consistent.

Additionally StarWind solution works with any hypervisor that runs on existing general-purpose hardware or even in private or public Cloud (e.g. Amazon or Azure). This solution will not affect performance of the whole system since the DR site does not require high performance. Our snapshots technology restores to the latest integral point of data even if corruption occurred on the primary and DR sites.

More information about StarWind Asynchronous Replication solution you can find on our web site, by following this link: https://www.starwindsoftware.com/asynch ... eplication

I hope that helped!

[epalombizio]

Is Starwind application aware? If SQL was run on a VM running off of Starwind, How would Starwind have any insight into SQL and to ensure DB activity is paused before taking a snapshot?

Thanks

[andrea.caldarone]

Hello Andrea,

Thanks you for sharing your question with StarWind community.

You have a guaranteed replication to the DR site, and the data blocks will be written to DR site in the same order so as they appear on the primary location. The trick here is that replication runs on the background, thus it doesn`t affect the primary storage tasks processing speed. Also If the data on the primary site gets corrupted for any reason (malware, human factor, etc.), then it will StarWind allows to recover to any point in time from snapshots located on DR, unlike traditional DR solutions. Obviously StarWind guarantees that data will be delivered and consistent.

Additionally StarWind solution works with any hypervisor that runs on existing general-purpose hardware or even in private or public Cloud (e.g. Amazon or Azure). This solution will not affect performance of the whole system since the DR site does not require high performance. Our snapshots technology restores to the latest integral point of data even if corruption occurred on the primary and DR sites.

More information about StarWind Asynchronous Replication solution you can find on our web site, by following this link: https://www.starwindsoftware.com/asynch ... eplication

I hope that helped!

Hello Anatoly,

I've not understood one crucial point, does StarWind works on snapshots to make asynchronous replication?
1) if yes, how much frequently I can take a snaphot? For example VMWare SRM that works on snapshots have a minimal RPO=15 minutes
2) if yes, are the snapshot taken with a VSS provider? Are application aware?
3) if no, if you use block level replication, do you have consistency group yes or no?

[Anatoly (staff)]

Yes, it is. But those are service snapshots, some kind of journals. LSFS constantly records the data changes. I hope that makes sense.
So, regarding your questions:
1. Taking snapshots is not mandatory to have DR copy. you can do it manually, or script the process with PoSh
2. Yes and no. Service snapshots are not application aware, but if you will take snapshot manually or script it - backup will be consistent.
3. Not relevant I guess