“The system is down.”
Whether you’re a major hospital, a cryptocurrency trading site, or an online travel agency, you probably fear hearing those words. The loss of important data, even for a few seconds, can be disastrous for your business. That’s why you probably have a disaster recovery site in place: as insurance against your system failing
In fact, as more companies continue to adopt the cloud for more and more mission-critical workloads, Disaster Recovery in the cloud is proving to be a valuable first use case. DR takes the overwhelming task of migrating mission-critical workloads into the cloud and somewhat minimizes it by taking live, production data out of the equation.
But even with this seemingly simpler use case, DBAs can face many hurdles. In this post, we’ll look at what these hurdles are and how to overcome them to successfully manage DR in the public cloud.
Hurdle #1: Performance
Data performance in the public cloud can be unpredictable. The cloud vendors set “throttles” to place limits on the speed and flow of data available through their shared, virtualized environment. These throttles keep your mission-critical databases and workloads from getting the same ultra-fast performance you had taken for granted on-prem. This can leave your end users with a very, very, VERY slow experience. How do you work around these throttles? By refactoring your databases and workloads for your cloud vendor of choice. But refactoring causes its own headaches since it is a time-consuming, risky, and costly process.
Hurdle #2: Scalability
To run your large, complex databases and workloads on the cloud, you need large VMs. But the cost of running these larger VMs can spiral out of control. Especially since resources such as compute power, data capacity, and performance are all tied together in public cloud architecture from a pricing standpoint. So, you need to overprovision one to get enough of another. This impacts your ability to scale in an economic fashion.
Hurdle #3: Licensing Costs
Databases such as Oracle and Microsoft SQL Server are licensable by CPU core. And if you’re trying to move one of these databases into your DR on the cloud, your budget can quickly get out of hand. Why? Because, as I mentioned before, customers must overprovision the number of vCPUs to get the amount of IOPS or throughput that they need for performance-hungry and latency sensitive applications.
In addition, replicating data from production to your DR on the cloud at the infrastructure level can also have a serious impact on your licensing costs by charging you for both the license(s) used in production and license(s) used in DR.
Silk – the DR in the Cloud Supercharger
So what’s the best way to overcome these hurdles and successfully kickstart your DR in the Cloud use case?
The Silk Platform lives between customers’ databases and the underlying infrastructure. By separating the two, it is easy to lift and shift databases from on-prem to the cloud and back again, making DR in the cloud a breeze.
Silk also offers 10x faster performance than native cloud platforms, meaning your back-up environment is sure to run as fast or better than on-prem. No worries there! The platform does this by decoupling performance from your cloud resources, so you aren’t limited by the throttles that the cloud vendors have put in place. Breaking the link between performance and resource also makes it easier to provision only as many VMs as you need for your DR system. Silk also makes it possible to dynamically scale your cloud resources should the need to failover to DR occur. Both capabilities means that Silk customers can cut their cloud bill by only paying for what they need, when they need it.
But that’s not the only way that Silk helps you keep more money in your pocket. The platform makes it possible to provision small compute instances while still meeting (or exceeding) your performance requirements. This has a drastic effect on your database licensing requirements, which in turn, effects your overall costs. During a failover, you may be able to transfer your database licenses from your no-longer-running production system. Meaning you can scale up compute resources without taking a hit from your database provider.
And if you’re planning to replicate database data from production to DR in the cloud at the infrastructure level? Depending on your choice of replication technology, Silk can offer you either a minimal target database VM or none, minimizing your exposure to extra core licensing costs.
Ready to get started with your DR on the cloud use case? Visit www.silk.us to learn more.