Today’s hybrid cloud and multicloud environments reflect the overarching reality that cloud usage – including all manner of software delivered as a service – is ubiquitous, so much so that it’s virtually invisible to most users.
That’s a good thing in terms of technology access and progress, but it comes with a potential downside: If you’re not paying attention, you might be in for some budget surprises.
Hybrid cloud costs: How to prevent wasteful spending
“The cloud has allowed organizations quick, almost unfettered access to a multitude of tools and capabilities,” says Manish Srivastava, VP and GM of IT asset management at ServiceNow. “This access can and will lead to waste during operations inside the cloud.”
The flexible, scalable, and diverse nature of hybrid cloud adds urgency to the need to keep a close eye on spending. We recently covered some common – and potentially expensive –misconceptions about hybrid cloud costs. Now let’s take a look at some expert tips for reducing and optimizing your cloud spending.
[ Learn the do's and don'ts of hybrid cloud: Get the free eBook, Hybrid Cloud Strategy for Dummies. ]
1. Watch hybrid cloud usage patterns and consider chargebacks
The first step: Don’t wait for a bloated bill to take action. Sure, surprise bills can happen to the best of them. They shouldn’t be the only thing that motivates you to analyze and optimize spending, however. This is as true now as ever, Srivastava says.
“Cost reduction must be a top priority and treated as a standard process, especially during COVID, when IT budgets are facing increased scrutiny,” Srivastava says. “It is important to monitor shifting usage patterns for various cloud applications as employees work from home and at some point in the future start coming back into offices. Companies can get started by rightsizing budgets, pre-buying compute, developing spending agreements, and terminating unused and unidentified spend.”
Gordon Haff, technology evangelist at Red Hat, points out that you need to have enforceable strategies in place to make sure departments and individual users don’t leave the lights on, so to speak. He points to chargebacks, whereby IT bills cloud spending back to teams or departments based on their usage, as one tactic for doing so. A developer might be less likely to accidentally leave a public cloud instance perpetually up and running if they know it’s coming out of their team’s budget, for example.
“Establish default policies and shut down services that aren’t being actively used,” Haff advises.
A thorough understanding of what you’re actually spending is fundamental to proactively managing hybrid cloud costs. The explanation for many unexpected costs begins with something along these lines: “We didn’t know about [X].”
“First, an organization needs to identify cloud spend,” Srivastava says. “This process fluctuates organization to organization, but it usually involves analyzing business, security, and technical information (owner, cost center, application, project, environment). The more dimensions available in the analysis, the better breakdowns and tracking an organization will have across cloud spend.”
Optimization grows out of that type of analysis, and again needs to be a codified part of your hybrid cloud strategy.
“Cost optimization needs to be viewed as a governance component and a form of standard operating procedures,” Srivastava says. “The faster businesses can understand their entire cloud landscape, clarify policies, and enact accountability, the better they will be at managing the wide variety of cloud spend sprawling throughout their business, particularly in this current work-from-home environment.”
[ Working on hybrid cloud strategy? Get the four-step hybrid cloud strategy checklist. ]
2. Set up alerts and other early warning systems
If your cloud service provider(s) offers tools for keeping tabs on what you’re spending, use them. It’s part of staying proactive, and some tools help automate certain types of responses to signals that your spending is moving out of its expected range.
“Use billing alert features,” Haff says. “In general, large cloud providers don’t offer hard cost caps but they do provide a variety of alert types – and often, the ability to create programmatic actions in response to alerts.”
Haff necessarily points out that your response to an alert isn’t likely going to be uniform: The next step(s) will depend on what’s driving the activity. Remember, the goal isn’t necessarily to cut spending across the board; it’s to cut unnecessary spending. It’s a distinction that sometimes gets lost in discussions about optimizing cloud costs.
“How you respond to usage (and therefore cost) spikes will depend on factors such as whether the service is business-critical or if it was just a data analysis experiment gone wrong,” Haff says.
3. Monitor, monitor, monitor
Chris Akritidis, COO at Netdata, casts another vote for using your providers’ native billing alerts, spending budgets, and other cost-management tools. The heterogeneous, distributed nature of hybrid cloud also means you’ll want to consider other ways to ensure visibility, and there are a variety of third-party tools for doing so.
“Go deeper and don’t wait for your provider to give you all the tools you need to be proactive,” Akritidis says. “Monitor everything you possibly can in real time so that you don’t have to wait for a specific cost element to rise enough to come to your attention.”
The type of analysis and optimization Srivastava describes above depends on this kind of visibility into the various components of your systems, whether in a public cloud or on-premises.
Srivastava notes that this is an area where machine learning will play a growing role, in concert with automation.
“The use of technologies such as ML [and] AI to identify behavioral patterns and propose spend optimization alternatives is key,” Srivastava says. “Machine learning can be leveraged for suggesting cloud resource tagging and spend anomaly detection as well. AI for hybrid cloud optimization may include proactive provisioning based on anticipated workloads, change control recommendations, or cost-effective image configurations, to name a few.”
[ What are the biggest benefits of an open hybrid cloud strategy? Learn more about Red Hat's point of view. ]
4. Make cost optimization a part of your design and architecture
From a budget standpoint, planning and building a hybrid cloud architecture is as much a financial tool as it is a technical strategy – or at least it should be. Data transfer between environments is a prime example. Not designing with egress costs – the price tag on moving data out of a cloud – in mind could result in soaring costs as well as other potential issues, such as latency.
[ Related read: Hybrid cloud costs: 5 misconceptions that can cost you money ]
“You can just start moving workloads to the cloud and reach back to the datacenter without really thinking carefully about architecture and design,” says Lenley Hensarling, chief strategy officer at Aerospike. “The key is where you put what part of the applications and the effect that has on where the data is being moved from and to.”
Running a microservices-based application in a hybrid cloud environment raises the stakes here. Hensarling and Haff both note that careful design and decision-making about what should run where is a significant part of managing your long-term costs.
Hensarling shares an example of a microservices-based customer engagement app running in a public cloud, for good reason.
“You get tremendous gains in elasticity as workloads change during the day or seasonally,” Hensarling says. “This pays real dividends in being able to scale as needed, but not carry the cost of infrastructure for peak loads.”
But what if you’re processing the invoices generated by that cloud app back in an ERP order system on-premises? Not accounting for this could drive up data costs, especially as the app scales. Proper planning and design can mitigate that, however, so that you can still achieve the benefits.
“You can cache most of the transactional data and the reference data up in the cloud,” Hensarling says. “Then, compress the data to drive the invoices and send the minimal amount of data needed to execute the required transactions in the back-office systems.”
This often boils down to architecture choices, especially when breaking up a larger system into microservices. Making these decisions with cost optimization in mind is important, especially when thinking in terms of data and how it flows.
“What data is kept where, and what data is actually required to execute transactions in other systems, needs to be carefully thought through,” Hensarling says.
[ Learn more about hybrid and multi-cloud workload strategy: Get the free eBook, Multicloud Portability for Dummies. ]