I still remember the week I realized that the cloud would most likely live up to its hype. It was a couple of years ago at the PASS Summit, and Microsoft had shifted their marketing message from being all-in on the cloud to a more pragmatic hybrid approach. In many of my offline interactions that week, I heard story after story of successes in moving some workloads to the cloud, resulting in significant cost savings and greater flexibility. That was the week that turned this skeptic into a believer that the cloud is the real deal, and it’s here to stay.
Although not every application is a perfect fit for cloud deployment, the fact is that cloud-based services are rapidly becoming the standard for new development. Platform as a Service (PaaS) vendors are also making it very easy to move existing on-premises applications to the cloud through myriad tools for data and metadata migration.
Benefits of Cloud-Based Database Applications
As the PaaS offerings from vendors such as Amazon and Microsoft have matured over the last few years, the benefits of moving workloads to the cloud have become clearer.
Decreased cost. Let’s say you decide to run your own standalone Microsoft SQL Server machine. Assuming you’re buying modest hardware, you’re going to be out $5,000 for the box, not to mention the periphery (network equipment, UPS units, etc.) required to set up a server and keep it connected and running. Then there’s SQL Server licensing, which for Standard Edition only was around $4,000 per core as of this writing. Assuming your modest server is a quad-core machine, you’re at $16,000 just for SQL Server licensing. If you plan to have redundant hardware to account for high availability or disaster recovery, expect to double (or more) these fixed costs. Also, keep in mind that servers don’t live forever; plan to repeat this purchase every 3-5 years. In contrast, the recurring cost model of PaaS applications reduces the steep capital expenditure into a more finely tuned pay-as-you-go economy, often resulting in lower total cost.
Time to market. Think about the things that will slow down your development or deployment timeline. It takes time – sometimes weeks or months – just to get approval for a capital expenditure for the hardware and software. After that, the new machine has to be built by the manufacturer, delivered, and set up in your data center. Install the operating system, configure it on your network, and then you can finally install SQL Server. In the absolute best case, this is a weeks-long process; in reality, going from zero to on-premises SQL Server will take a month, perhaps much longer. Building a PaaS database can be completed in minutes rather than weeks or months; in fact, earlier today I built a SQL Database in Azure in about the same length of time it took me to type out this paragraph. If time is a factor (and really, when is it not?), the speed at which a new cloud database can be created is a very compelling argument for using a PaaS architecture.
Managed services. If you’re hosting your own database server for a 24×7 application, your organization is on the hook for handling things like operating system hiccups, hardware failures, network snafus, and power outages. Keeping resources and staff to handle these issues takes time, training, and most importantly, money. Add to all of that the ongoing environmental costs of running a data center: cooling, power, backup power apparatus, and physical security, among others. By using a cloud platform to develop and run your database applications, you’ll shift the responsibility of data center management over to your cloud vendor, saving a great deal of effort and money. You’ll also rest a bit easier, since software updates and service SLAs will also be handled by the PaaS host.
Easy scalability. With your on-premises database server, what happens in six months when you discover that the hardware can no longer keep up with demand? You could upgrade the server hardware (assuming it’s got headroom for that) or just buy a new larger server. Similarly, I’ve seen a few cases of buyer’s remorse where the needs were overstated and the server hardware was much more robust (and therefore, more expensive) than it needed to be. Overbuying or underbuying hardware is always costly. Conversely, when building a database application in the cloud, it is relatively easy to set – and later change, if needed – the resources allocated to it. As the processing and storage needs change, the resources can be scaled up or down in seconds with just a few mouse clicks.
Conclusion
Platform as a Service offerings have evolved and improved significantly in the past couple of years. These cloud services make a strong argument for cost savings, time, and flexibility for both new development and migrated applications. As the volume of customers migrating to the cloud continues to increase, prices will continue to fall, furthering the cost benefit advantage of cloud-based applications over their on-premises counterparts.
