When we talk about “the cloud”, we’re actually talking about not just one cloud, but different kinds of clouds. Enterprises looking to move their business processes to the cloud have essentially three categories to choose from: private (i.e. your very own cloud), public (i.e. Amazon Web Services) and a hybrid of the two (also called “multi-cloud”).
For some time now, enterprises have been leaning heavily towards the hybrid option, for a variety of reasons. For a start, it reduces dependence on a single cloud provider. Hybrid clouds also enable greater flexibility in terms of which apps go where, depending on criteria like performance, cost containment and cost optimization. Multi-cloud solutions are also much more resilient during outages and maintenance. And a hybrid approach enables enterprises to take advantage of advanced capabilities that may be offered by one cloud provider but not others (think about IBM SoftLayer and Watson, for example).
“The question for all IT transformation strategies is when to use public cloud services, when to use on premise capabilities and how to govern both as one. As speed accelerates and complexity increases nothing should be handled implicitly, manually or as an afterthought ” says Geoff Hollingworth, Head of Product Marketing Cloud Systems at Ericsson.
See Also
How to get the multi-cloud business advantage without sacrificing security or control
However, as you might expect, there’s a lot more to hybrid cloud than simply bolting multiple clouds together. In fact, says Derek Collison, founder and CEO of Apcera (majority-owned by Ericsson and part of the company’s cloud offerings), it creates some formidable challenges, especially in terms of orchestration and governance for apps.
“A single enterprise organization will have a whole host of languages, frameworks and software architectures in flight at any one time powering their applications and their business,” Collison explains. “Engineering an infrastructure and network configuration that can support the wide variety of requirements and diverse workloads is one of the hardest parts of doing this at scale with the requisite amount of security and performance.”
Consequently, he continues, enterprises need a programmable network and a multi-cloud platform that is enforceable, visible and can move at the speed of the developers and the DevOps layer. And that means a platform that is designed with governance built in at the core, rather than multiple solutions bolted on after vulnerabilities or risks are exposed.
“This lets the business and the IT operations group set policies or rules to control the network access to, from and within the system while still allowing the users or developers to configure the resources and connectivity they need without any manual intervention,” Collison says.
He adds that the orchestration capabilities of such a platform are essential when deploying and running vast numbers of new, loosely coupled services and workloads that can span multiple clouds. “Ideally, the governance platform needs to be capable of orchestrating any workload on and across any cloud – public, private, or hybrid.”