Is cloud bursting the best solution for compute capacity overload?

ptnphotof - Fotolia

Is a cloud bursting architecture difficult to implement?

Cloud bursting helps organizations use the public cloud to manage sudden spikes in demand. But what challenges or issues might it introduce?

Cloud bursting is the process of moving workloads between a private cloud and a public cloud to accommodate changing workload demands.

For example, if you have a spike in traffic to a website and need to increase the number of servers in a cluster -- or you are running an unusually heavy load on a service and need to create more instances of that service -- you can use a cloud bursting architecture. A private cloud can handle these kinds of demands alone, as long as you have the hardware to support additional virtual machines. However, when you run out of infrastructure in your private cloud, cloud bursting lets you use the public cloud to expand your available resources.

While cloud bursting is a big benefit of the hybrid cloud model, it's difficult to implement. As a result, it's important to determine if the need for a cloud bursting is sufficient enough to warrant taking on these implementation challenges.

Cloud bursting architecture brings new IT challenges

One common challenge when implementing a cloud bursting architecture is that private and public clouds may run on different cloud platforms. Determine how well your configuration scripts work in both public and private clouds. You might need, for example, one set of configuration scripts for your OpenStack-based private cloud and another set to burst into Amazon Web Services. However, one way to avoid this is to use a service that abstracts the cloud implementation details. For example, you could use the RightScale cloud management service to abstracts processes, such as creating a virtual server. Alternatively, you may be able to use hardware-based load-balancing across clouds with a product like F5 BIG-IP Local Traffic Manager.

While cloud bursting is a big benefit of the hybrid cloud model, it's difficult to implement.

Even if you establish a cloud bursting architecture, consider the impact of running a single service in multiple clouds. Network latency, for example, can be an issue unless you host your private cloud in the same data center as the public cloud to which you are bursting. Watch for egress charges if you are transferring large amounts of data out of a public cloud and into a private cloud.

Also, consider the location of the data that your cloud service needs. Questions to ask include:

  • Are data copies available in both clouds?
  • Will you have to copy data during the bursting process to ensure the public cloud has the most up-to-date version?
  • Are there constraints on how you can manipulate the data?
  • Will having multiple data copies introduce inconsistencies if both copies are updated during the bursting operation?

Despite its benefits, a cloud bursting architecture is not always easy to implement. You can also consider alternative approaches to managing workloads, such as using the public cloud to host services that have spikes in demand that exceed private cloud resources.

Next Steps

Explore the role of DR and cloud bursting in hybrid cloud

Planning for a private cloud project

Is a multicloud model right for your business?

Dig Deeper on Cloud architecture design and planning