The big promise of the hybrid cloud is the efficiency it can bring to data centers. Most data centers were initially built for specific tasks, but now they are taking on more responsibilities as IT has become essential for business, whether it’s running complex analytics, providing storage, protecting sensitive data, and more. As a result, modern data centers have become as messy as a garage filled with boxes and tools.
Moving storage and compute resources to the public cloud can free up space in the data center for other tasks that might be more important for specific business goals. That said, the ability to move your persistent storage and computing resources into the cloud may solve certain problems, like the need to buy and store multiple servers, but it doesn’t solve every storage and/or computing challenge. What if you’re in a highly seasonal business, or you provide a service that experiences regular but unpredictable spikes in activity? In these cases, you may not know in advance how much storage you need, no matter where it lives.
One of the most buzzed about applications for the hybrid cloud has been cloud bursting. In this article, we’ll explore what cloud bursting is, when you should consider it, and what challenges you should expect in implementing it.
The Storage Problem
Let’s say you have an application that runs in your private cloud, like an e-commerce site. For 11 months out of the year, your existing computing and storage resources may be sufficient, but during the holiday shopping season, you might find yourself in need of additional capacity.
This situation can present a real catch-22 for any business. You can either invest in expensive new computing power that will go unused for most of the year, resulting in an inefficient use of resources, or you can underprovision, and experience lost traffic—and revenue—if you have downtime during high-demand periods.
A Hybrid Cloud Solution
Cloud bursting offers a compelling solution to this long-standing problem. In essence, cloud bursting allows an organization to rent additional storage and compute resources on demand. In theory, this is a win-win arrangement, enabling organizations to avoid embarrassing and costly downtime without spending money on resources that they don’t need for most of the year.
Technically, this can be accomplished a couple of different ways. An organization might run an application locally and then “burst” another instance onto the public cloud during peak demand. Alternatively, the entire application can be moved to the public cloud to free up local resources for other applications. Which approach is best depends on several considerations: Is the application configured to support multiple instances running at the same time? Can the application handle the increased latency that’s likely to occur when running entirely from the public cloud?
To Burst or Not to Burst?
The concept of cloud bursting (also called workload overflow processing) has been around for several years, but has yet to really catch on, particularly at the enterprise level. Why is that?
The main reason is that cloud bursting is a simple concept that’s technically difficult to accomplish at both the infrastructure and application levels. Firstly, it’s difficult to move storage to and from the cloud. Bandwidth limitations represent a real, practical constraint that limits the ability to move resources to and from the cloud. On top of that, different environments aren’t always compatible. If your private cloud and public cloud are set up to handle very different responsibilities (which is common), then you may have a lot of trouble setting up cloud bursting.
Secondly, software must be configured to run multiple instances simultaneously. If you’re building a new application from scratch, this can be engineered in. If you’re working with an already extant application (which most people are), then it can be a difficult and time-consuming process to add this functionality. The more complex your application and the more services it interacts with, the more difficult it is to burst.
Given these limitations, cloud bursting is usually recommended for high-performance applications that won’t suffer from the latency issues that come with moving between the public and private cloud. Additionally, you should consider whether or not the application handles data that’s subject to regulatory mandates or compliance standards like HIPAA or PCI DSS.
Services That Can Help
While cloud bursting remains a significant technical challenge, a number of options have emerged in recent years to help address some of these issues. Bracket Computing and VMTurbo both offer solutions that work by abstracting various parts of the cloud provisioning and management process, removing a lot of the complexity associated with setting up and managing cloud bursting. VMTurbo emphasizes automating critical tasks like load balancing and scaling, meaning it can figure out when it’s cheaper to run a given workload on the public cloud, execute the burst, and then bring it back. Bracket is designed for financial institutions, emphasizing unified IT security controls, visibility, and the ability to cloud burst using sensitive data.
Additionally, both Amazon Web Services (AWS) and Microsoft Azure claim that their infrastructure complies with HIPAA, PCI DSS, and many other compliance standards, and both have taken steps to reduce downtime, thereby alleviating two of the major infrastructure concerns associated with hybrid cloud setups in general and cloud bursting in particular.
The Importance of Planning
Before you consider cloud bursting as a workload overflow solution, you’re going to want to spend some time thinking about your particular business needs. A cloud server architect can help you figure out whether cloud bursting makes sense on your current environment, and show you what specific steps you’ll need to take to get it up and running. After that, you’ll most likely want a dedicated dev ops engineer with experience in hybrid cloud systems to make sure you’re achieving the efficiencies that the cloud promises.