If you are a Microsoft customer looking at how to migrate Microsoft SQL Server from on-prem to Azure, this blog is for you. We will share 5 considerations for achieving high database performance while avoiding any pitfalls related to refactoring or over-spending on cloud bills.
#1: Know the Bounds of Database Performance
First, it’s helpful understand the limits of database infrastructure. SQL database workloads requiring more than 2 GB/s throughput or 160k IOPS are not able to achieve this level of performance on Azure without some help on the data layer. Azure Ultra SSD maxes out at this threshold. While these numbers may be perfectly fine for a great number of applications, they may not be enough for transactional or analytic database applications on Microsoft SQL Server. These applications would need to be refactored or left on-prem. We’ll discuss how to overcome this performance limitation by choosing the right data platform on top of Azure.
#2: What’s Your Block Size?
The changing nature of Microsoft SQL Server workloads (different block sizes, changing demand) can create a non-consistent SLA with inconsistent latency in the public cloud. This happens when your applications combine transactional and analytic workloads. Operations using large blocks can dramatically affect performance.
Employing a data platform on top of Azure that delivers consistently high performance with low latency that is independent of read/write block size or pattern will shield your database performance from variable workloads and neighbors on the network sharing resources.
#3: Should You Refactor or Lift and Shift?
Migrating Microsoft SQL Server workloads to the Azure cloud is a natural choice. Microsoft offers great incentives to stay on the platform and leverage the skillset of your team as you expand your use of cloud services. You can continue using the Microsoft SQL Server database on Azure, so there is less risk and refactoring involved that goes in to changing databases. But if you have a large database that requires high performance (see consideration #1 above), you’re going to need to adapt that database to the performance limitations of the cloud.
#4: Consider Your Data Footprint
For maximum efficiency, it is recommended that you take advantage of data reduction services, including deduplication, data compression, and thin provisioning to reduce your data footprint and keep cloud costs in check. While cloud infrastructure does not offer data reduction services, you can get this capability by integrating a data platform or tool that does offer it.
#5: Keep Track of Database Clones
Database clones are frequently used in the CI/CD pipeline for Dev/Test, analytics, and production. On-prem Microsoft users may be surprised to find that creating writeable clones in the cloud requires the creation of full clones that take up capacity and can be time consuming to manage. Our advice is to keep track of clones to make sure they are not impacting performance or contributing to an unexpected cloud bill.
Using Silk to Supercharge Microsoft SQL Server on the Cloud
With these 5 considerations in mind when migrating Microsoft SQL Server to Azure, we’d like to introduce you to Silk, the database supercharger.
The Silk Platform is a data layer that sits invisibly between your databases and cloud infrastructure to supercharge your databases. With Silk’s architecture, there is virtually no limit on throughput, IOPS, or disk size. Data flows seamlessly from cloud to cloud and back to on-prem if needed. You can dynamically scale performance up and down automatically or manually as workloads change.
Silk’s proven performance makes it possible to:
- Load more than 20 million rows in 1 second,
- Load more than 7TB of data into a SQL database in less than an hour,
- Scan a 1GB data table in 0.5 minutes.
That’s the performance of Microsoft SQL Server on-prem but also in the cloud. You can easily run the most demanding SQL workloads on Azure and refactor on your own timetable.
Learn more about the Silk Platform and download our data sheet: The Boost in Performance Your Databases Need: The Silk Platform on Microsoft Azure.
