In the enterprise market, there seems to be a growing consensus that two clouds — or even more — are better than one. Learn more about hybrid cloud markets in this article from TechTarget featuring Opex Technologies’ president Auburn Holbrook.
For reasons ranging from compliance to disaster recovery, an increasing number of organizations are choosing to extend their IT environments across private and public clouds. And as they do, the hybrid cloud market will soar from $25.28 billion in 2014 to $84.67 billion in 2019, according to estimates from research firm MarketsandMarkets.
But before they reap hybrid cloud benefits, businesses must clear a number of operational and management hurdles. Among other tough decisions, they must determine which workloads to run in the public cloud and which to run privately, how to optimize their network for a hybrid IT model, and, of course, how to manage and minimize costs.
Allocate resources to maximize hybrid cloud benefits
One of the biggest hybrid cloud benefits is the flexibility it gives organizations, allowing them to place some applications in the public cloud, while others remain on premises. It’s a best-of-both-worlds kind of model, said Pamela Wise-Martinez, a cloud and enterprise data architect at Pension Benefit Guaranty Corp., a U.S. government agency that manages pension plans.
“[Our data] is very personal, it’s very private and it’s very important for us to protect it, so we’re never going to go 100% [public cloud],” Wise-Martinez said. “But what I can see — and this is why hybrid cloud is so important, and I think that’s why it makes sense for most government agencies — is you’re going to have some on-premises and off-premises services.”
And while that flexibility is a boon to many IT shops, it also presents a challenge — namely, deciding which workloads to run where. For many organizations — especially those in vertical markets like healthcare and financial services — governance and compliance needs still drive that decision, said Auburn Holbrook, president of Opex Technologies, a cloud advisory and brokerage firm based in North Carolina.
“[Some organizations] are still hesitant to put certain data sets out in the hyperscalers,” Holbrook said, noting Amazon Web Services (AWS) and Google as examples of these hyperscale cloud platforms. “As those compliance needs are being met [in the public cloud], I’m sure they will be less hesitant. But, for now, it’s too hard to audit. It’s easier if they can separate [data] and keep some on premises.”
In other instances, the age or type of an application dictates whether it should run in the public or private cloud, said Karl Scott, executive consultant at W. Capra Consulting Group, an IT consulting firm based in Chicago. For example, legacy applications that would require significant refactoring to run efficiently in the public cloud are often better left on premises or in a private cloud.
“The challenge most organizations have when trying to take the older applications and put them in the public cloud is, quite frankly, that they consume more resources than are really needed,” Scott said. “Because they are older applications and they are kind of bloated, the cost savings are sort of eaten up.”
Once an organization decides which applications to place where, it’s crucial for it to have a “clear integration or application architecture strategy” in place for hybrid cloud, Scott added. To do this, they should implement more modern application integration methods, such as the use of RESTful APIs.
Optimizing hybrid cloud performance and costs
Just as organizations need to prepare some applications for hybrid cloud, they must also prepare their networks for the shift. Without understanding the new data traffic patterns that occur within hybrid cloud — a model in which data often moves across different IT environments — it’s difficult to ensure application performance and resiliency, Scott said.
“As opposed to everything coming back to a central data center, you have services going to, for example, a private cloud that is hosted, say, by CenturyLink, and then you might have public-facing services that you are hosting out in AWS,” Scott said.
To support these new traffic patterns, and maximize hybrid cloud benefits, organizations should look into options such as multipathing, which creates multiple physical routes for data to travel between cloud environments. This can reduce latency, assist with load balancing and help maintain uptime.
In addition, organizations should consider establishing a dedicated network connection from their private clouds to their public cloud providers, Opex’s Holbrook said. While some public cloud providers offer these dedicated connections themselves — such as AWS’ Direct Connect service or Azure’s ExpressRoute — there are also third-party options from networking vendors like Level 3.
“There are some very cost-effective means of directly connecting existing computing environments, whether it’s within a customer-owned facility or an existing outsourced provider [to the public cloud],” Holbrook said.
Regardless of how they architect their hybrid cloud environments, all businesses should take proper measures to ensure they can minimize and manage costs. To do that, IT teams need to prevent users from getting carried away when spinning up new resources in the public cloud, Holbrook said.
“A development organization might spin resources up, then leave for the weekend, and those resources continue to run, even though they aren’t being utilized,” Holbrook said. “A lot of organizations’ cost structure is getting out of control, because they are not effectively managing that process.”
Organizations should also use cloud cost management tools to ensure they run only the cloud resource they need — and no more. While public cloud providers like AWS and Azure offer their own cost management features, users should supplement those with third-party tools, such as Cloudability and CloudCheckr, which tend to be more thorough.
Lastly, organizations should consider designating a single point of contact within their organization to track and manage cloud spending, Scott advised.
“It makes sure somebody is actually governing all of the cloud spend, and that somebody is actually monitoring and managing that on a monthly basis,” he said