Categories: Pure Storage
Catalogic is excited about showcasing our application-aware snapshot and replication solutions at the upcoming Pure //Accelerate 2017 show next week. Join us at Booth #2 and meet our technical team to discuss how we can add value to your environments or your customers.
The focus of this blog will be on some specific capabilities within Catalogic that allow you to maximize your Pure Storage investments, particularly around SQL Server (though much of this applies to other databases we support as well). We often see customers try to automate some of the snapshots and cloning through scripts, but scripts suffer from a variety of problems:
I cannot emphasize this enough. The enterprise risk you are exposing your organizations to is exactly the kind of risk that results in lost jobs when things go wonky. This is also the primary reason for data sprawl. Clean-up logic is almost always an after-thought with scripts, and copies often get made but never deleted.
Crash consistency is not application consistency. Application consistency is not application awareness. Being able to integrate with the SQL stack and the server stack seamlessly is important to deliver a reliable mechanism to deliver copies.
How many scripts still use important passwords in plain text? Don’t even get started on AD integration and RBAC.
Running these scripts on an enterprise FlashArray from Pure Storage is like buying a Formula1 car and saving money by fitting it with 4 bald tires. Penny wise, pound foolish. In fact, we’ve found so many concerns with scripts that we’re even giving away “No More Scripts!” buttons at //Accelerate.
Drop by to grab your button. We’ll be randomly selecting folks wearing the buttons to win one of the many remote-control cars we’re giving away.
Other than addressing these blatant issues with promoting and using scripts, what are some of the scenarios where Catalogic can really help deliver a better outcome for your organization, especially in regards to Microsoft SQL Server?
Disk Signatures are thumbprints that Microsoft uses to identify disks and it helps them deal with knowing what physical disks belong to which volume during boot time, as well as when it sees new devices. Windows Clustering especially relies on this mechanism to ensure it avoids “split-brain” scenarios for shared drives. When you want to restore a clone of your DB back to a cluster, the volume will show up with the same signature and can confuse Windows and Clustering. This is why scripts often fail to restore back to the original cluster, unless the original shared volumes are destroyed.
Catalogic will allow you to define a proxy SQL server that can be used to re-signature the disks before we mount it back to the original clusters. One of our Catalogeeks, @Martin Phan, explains this in a demo video.
Pure Storage provides a hardware VSS provider for physical SQL workloads, but this provider only works for VSS consistent snapshots on the local array. This provider predates Protection Groups and there is no easy way to get an application consistent copy over to your replica array. Some script-based solutions use a workaround that takes a snap, creates a new volume off it and replicates it. The problem with this approach is the amount of data that needs to be transferred every time – it’s the full copy, which is a colossal waste of bandwidth. On top of it, the clean-up logic is inefficient and can create sprawl within your organization. This is one reason why “snapshot assisted backups” from some of the new backup appliances are popular among Pure customers, because they can at least ship only the incremental changes, even if it means losing the compression/deduplication savings in-flight.
The most efficient solution requires building a VSS hardware provider that integrates with Pure Storage’s Protection Groups. Catalogic delivers that. We’ve built a hardware VSS provider for SQL that also handles replication. It does one better than purpose-built snap assisted backup solutions by moving data at the storage layer, thereby preserving the 5:1 data reduction Pure Storage customers are getting. As a recent joint Pure/Catalogic customer told us, “A storage dollar not spent on Pure is a dollar wasted.” We agree! Don’t buy low performant backup appliances just to get an app-consistent offsite copy.
AlwaysON Availability Groups from Microsoft SQL Server have been a revelation for many users that can’t afford much downtime. The swift adoption of AlwaysON shows that HA configurations are a core requirement in most companies. However, establishing an availability group across multiple sites may be a concern. It requires sending all the data over the wire and it requires the Pure Storage array to have to work on deduplication and compression of the same blocks again. We see customers often use replica DBs for testing and after every test, reseeding those databases can be a pain for the administrators and places a lot of burden for your network and storage to achieve a net zero effect.
Catalogic has an instant seeding option for remote sites. Using it, you can reseed the replica database from an existing snapshot in the replica site. Any unhealthy/uninitiated replica copy can be recovered and brought back in synch with its primary database with little to no additional data transfer over the wire. Pure Storage also doesn’t have to spin its wheels trying to compress these blocks. This is much quicker, requires less network bandwidth and is more efficient on the storage.
This is just a small sampling of the unique SQL Server features built into the Catalogic Copy Data Management platform to allow you to leverage and maximize your investments in Pure Storage. We would welcome the opportunity to interact with your teams and discuss how we can help improve your processes around:
Drop by to meet us at Booth #2 next week, where we’ll be able to show you all these features via live product demos. Or feel free to email us anytime at Pure@Catalogicsoftware.com.