Getting to Hybrid in a Multi-Cloud World

Public clouds can be a blessing and a curse. They’re easy to use in limited ways, but hard to turn into a strategic resource. To make them an effective part of an enterprise infrastructure, there’s work that has to be done and it’s not insubstantial.

Most enterprises are using multiple cloud providers through relationships that look more like accidents than careful planning. A development team started using one provider, while a partner had resources in another. A mobile app had its back end in a third, and suddenly you’re in a multi-cloud situation.

Managing multiple clouds comes with a set of complexities that can be dealt with by getting to a more effective hybrid management arrangement. Hybrid architectures coordinate operations across clouds and can be more efficient and effective than multiple, disconnected clouds. The most recent 451 Research Voice of the Enterprise, Cloud Transformation study shows how enthusiastic support for hybrid deployment models for new infrastructure has become. Respondents identified hybrid cloud as the fastest growing (at over 72%) area of expected application deployment over the next two years.

To get to hybrid cloud capabilities, you’ve got to be able to coordinate and integrate operations across cloud providers. It can be a challenging task, as the controls and even billing models can differ significantly between clouds. One of the largest challenges starts at the beginning of the relationship with a provider: normalizing their contract and terms to align with business requirements. The nature of cloud services has meant that there are few options that can be negotiated with major providers. That leaves organizations with the burden of determining how to match their internal legal needs. Data ownership, termination details and other aspects may necessitate the use of compensating controls, such as data encryption.

If you’re formally integrating a cloud partner, you’ll need to ensure that a sufficient network interconnect is in place to deliver the performance that the workloads you’re placing there will need. That will probably be more than the Internet VPN that is most provider’s default. This will link your existing resources in datacenters or colocation to the cloud provider. Some colocation providers offer exchange services which can be sufficient for many use cases. They’re typically easy to set up and are arranged through the colocation provider. Datacenter connections may require working with a communications service provider and using cloud provider access services, such as Amazon’s AWS Direct Connect, Google Cloud Platform’s Dedicated Interconnect or Microsoft Azure’s Express Route.

The largest part of making a cloud partner part of your hybrid infrastructure is centered around workload deployment and operations mechanisms. Existing processes and tooling have to be adapted to the provider’s environment. The amount of work required will depend on the current level of operational maturity that an organization has achieved. If you have orchestration systems that abstract operational controls across different environments and retarget workload images, the work could be reduced to a process of adapting tools to the new cloud API’s. But that’s not the situation for the majority of enterprises. For these organizations, build scripts will have to be adapted and internal processes changed to deploy successfully in the new environment.

While there are cloud brokering services and some vendors, including Red Hat and VMware, offer bridges to public cloud, organizations are going to need to be able to build hybrid environments themselves, if they’re to gain the full advantage of the benefits of hybrid cloud and the independence that it can bring.

Even for those that have more mature environments, there is still the task of determining policies for workload placement. This starts with instance size alignments, making sure that equivalent capacities are matched up in old and new environments. There will also need to be connectivity and compliance alignment, to ensure that sensitive data winds up only where the right controls are in place.

While the transition to fully hybrid operation is complex, it’s a necessary step towards greater infrastructure flexibility. At the ONUG spring conference, we’ll be exploring the various aspects of taking on and integrating a cloud provider in the Hybrid Multi-Cloud working group session. We’ll look at ways to model costs and lay out integration tasks. There will be discussions of the working group’s progress and the Proof of Concept model for hybrid cloud. Join us to learn more!

Interested in more technology insights? Join the 451 Alliance to gain exclusive access to IT industry trend data.

Author's Bio

Eric Hanselman

Chief Analyst, 451 Research