Many IT leaders are reconsidering their approach to cloud. You may sometimes hear this process called “cloud repatriation”, but that terminology isn’t entirely accurate. With only a few exceptions, enterprises aren’t ditching the cloud altogether. Instead, they’re evaluating individual workloads on a strategic, case-by-case basis to see which ones might be better suited to run outside the cloud. That’s why the term “cloud rebalancing” better describes what’s actually happening.
This rebalancing process is an important part of your organization’s wider hybrid multicloud strategy. It’s about finding where each workload belongs and then finding the best way to get it there. Since each workload has its own unique requirements, there are no easy answers when it comes to rebalancing your cloud infrastructure. You’ll only know for sure after you consider all the possible angles.
To help you make more informed decisions during the cloud rebalancing process, we’ve put together the following list of seven questions to ask yourself.
1. Is cloud infrastructure costing more than expected?
Many organizations have learned the hard way that cloud isn’t always the most cost-efficient infrastructure option. There are several factors that drive high cloud costs, starting with raw infrastructure capacity for compute, storage and networking.
Compute often constitutes the biggest portion of the bill. This includes the cost of the virtual machines or containers used to run workloads, as well as accelerators like GPUs. The emergence of compute-intensive technologies like AI could drive up costs even further if you aren’t careful.
Cloud storage is also a big line item, and the more data you store, the more you pay. That means this expense will continue growing indefinitely as your workloads grow. You’ll have to pay both the cost of the storage capacity itself, and the cost of the services used to manage and access data.
Finally, cloud networking includes costs for data transfers and the services used to connect to a cloud or manage cloud traffic. In particular, data egress fees could drive up costs significantly if organizations are frequently moving data out of the cloud.
2. Do your workloads have existing cloud dependencies?
The more dependent a workload is on services from a particular cloud provider, the harder it will be to move off that cloud. It would require standing up equivalent capabilities on-premises or redesigning the application to do without those capabilities. Both of these options would likely prove impractical and cost-prohibitive.
Even if you’re able to replicate these dependencies on-premises, doing so won’t necessarily be the right move. You may decide it’s not worth losing cloud benefits such as automatic updates, access to new features and ease of scalability.
3. What’s your resource utilization rate?
It’s important to closely examine how your applications use specific resources like CPU capacity, memory and bandwidth. Resource utilization impacts performance as well as cost. You need to ensure your applications have the resources to run at their full potential, but you also need to avoid paying for more than they need.
Additionally, cloud spend can become unwieldy over time. Cost overruns are common, as seemingly innocuous expenses can add up unexpectedly. These “micro costs” include things like API calls and service features that start out inexpensive or even free of charge. They’ll inevitably cost more as workloads grow and the clouds change their pricing models. Because of this, it’s important to keep a close watch on all resource utilization, not just the main culprits (compute, storage, network egress).
To strike the right balance, you may need a hardware configuration that’s fine-tuned for the specific requirements of your application. You need to determine if you can get this kind of custom configuration in your current cloud environment. If not, that may be a good reason to move the application.
4. How valuable is cloud scalability for your workloads?
In addition to optimizing your current resource utilization, you need to plan for future growth. It’s true that the cloud provides scalability to support growing workloads, but it comes at a cost: Running bigger workloads logically means spending more on infrastructure capacity.
It’s also important to consider how a workload is growing. If growth is consistent and predictable, then you may not need on-demand cloud scalability. You’d be able to plan your on-premises infrastructure in advance, which would likely be the more cost-efficient option. In contrast, if a workload is growing unpredictably or spiking temporarily at specific times (like an e-commerce application during the holiday shopping season), then leaving it in the cloud might be for the best.
5. How does cloud impact your compliance requirements?
If you operate in a heavily regulated industry, such as healthcare, finance or government, then you’ll need to consider compliance as you decide where to host your applications and data. The same is true if you operate in jurisdictions with strict data sovereignty requirements. To meet these requirements, you’ll need infrastructure within specific borders, and you may not be able to get that if you rely exclusively on public clouds.
Cloud providers offer various industry-tailored compliance services—for a fee. A do-it-yourself approach to compliance could be more cost-efficient, but you’d need to ensure that you’re prepared to take control and execute the compliance solution you need.
Your industry may also have availability or resiliency requirements. These dictate the level of redundancy you’ll need to implement for your workloads. As you make your design decisions, you’ll need to consider how to meet these requirements. You’ll also need to account for the cost of adding the necessary infrastructure.
6. Does the cloud meet your uptime requirements?
Before you decide whether to leave a workload in the cloud, you need to stop and read the fine print on the cloud provider’s service level agreement (SLA). This can help you determine whether your current cloud environment is the best place to meet your workload’s uptime requirements.
Consider these points:
- What does the SLA cover, and what doesn’t it cover?
- Has the provider historically been able to meet the SLA?
- How would you be compensated if they didn’t meet the SLA?
- Where does the risk reside in the event of an outage, and what is the shared responsibility model?
- And perhaps most importantly, could you get comparable uptime and better manage risk using alternative infrastructure?
Only after you’ve answered these questions can you make an informed decision about whether your current cloud environment is the most reliable infrastructure option for your workload.
7. If not the cloud, then what?
Suppose you’ve considered all of the factors outlined above, and you’ve concluded that rebalancing some workloads off the cloud is the right move. That still leaves the question of where to host the workloads instead of the cloud. You need an alternative that can meet your performance requirements, ensure reliability and enable flexible infrastructure. On top of all this, the alternative needs to provide cost benefits that make migrating worthwhile.
To deliver all these benefits and more, an Equinix IBX® colocation data center could be the answer you’re looking for. In an Equinix data center, you get the best of both worlds for your workloads: the control of a private infrastructure environment paired with cloud-like flexibility. For instance, you can maintain data in a private storage environment instead of in the cloud. This allows you to temporarily move copies of your data into the cloud when the need arises, without having to pay egress fees to get that data back out.
Equinix can also help you design your infrastructure environment for optimal data center density. This puts you in a better position to achieve the cost and performance goals that inspired you to pursue cloud rebalancing in the first place.
You’ll also get access to Equinix Fabric®, our virtual networking solution. With Equinix Fabric, you can quickly set up the virtual connections you need to move your data and workloads anywhere they need to go. Regardless of which workloads you choose to move off the cloud and which you choose to leave in place, they can all run together on the same interconnected hybrid multicloud platform.
To learn how the right networking solutions can help you optimize your digital infrastructure across both cloud and on-premises environments, read the white paper Thriving with a hybrid multicloud strategy.
Subscribe to the Equinix Blog