Hybrid Cloud Computing Faces Multiple ChallengesBy Michael Vizard | Posted 02-22-2013
Hybrid Cloud Computing Faces Multiple Challenges
By Michael Vizard
Most existing examples of hybrid cloud computing are not all that complicated. They generally involve integrating a CRM application, such as Salesforce.com running in the cloud, with, say, an ERP application from Oracle or SAP running on-premise.
That type of hybrid cloud computing is generally not much more complex than trying to integrate two applications running on-premise; the only difference is that one of those applications is running in a third-party data center that needs to be accessed over a wide area network that is inherently latency sensitive.
According to Rebecca Wettemann, vice president of Nucleus Research, the return on investment in cloud applications is too compelling to ignore. Wettemann says Nucleus research finds that cloud applications deliver 1.7 times more return on investment on average over on-premise applications. At the same time, however, managing applications running on-premise and in the cloud is not easy.
“Integration is so important, which is why this is such a hot area of investment,” says Wettemann. “Still, it can be a pretty scary thing to do.”
When most vendors talk about hybrid cloud computing, their aspirations tend to go well beyond the comfort level that most IT organizations have with cloud computing today. IBM, for example, likes to paint a cloud computing picture that is application workload-centric.
“Cloud computing will be all about running the right application workload on the right system at the right time,” says Hector Hernandez, cloud computing business development leader for IBM.
Major elements of that IBM vision, however, will give most IT organizations a moment for deep pause. The idea that virtual machines carrying workloads that will be processed on systems that are best optimized for that workload raises a specter of complexity that most IT organizations don’ t have the tools and processes to manage. In fact, most IT organizations are more interested in how to stabilize an increasingly complex IT environment that is already made up of more static virtual machines than they can sometimes handle.
“Managing IT is tough enough as it is,” says Rich Hathaway, principal of RLH Telecom Solutions, an IT consulting firm. “Nobody is voluntarily doing anything like that today.”
Before most IT organizations get anywhere near that level of hybrid cloud computing complexity, there are a number of issues that need to be addressed, not the least of which is IT automation. Unfortunately, most of the management processes being used inside IT are either manual or use custom scripts that don’t scale well.
Worse yet, the overall virtual machine environment is becoming more heterogeneous. Startup companies such as Ravello Systems are addressing today’s complexity by making it simpler to take an application workload running on one virtual machine, such as VMware, and deploy on a cloud computing platform that supports Xen or KVM hypervisors. Currently available in beta, the HVX cloud hypervisor is a nested hypervisor that normalizes various cloud computing platforms.
Hybrid Cloud Computing Faces Multiple Challenges
According to Navin Thadani, senior vice president of product management and marketing for Ravello Systems, this approach allows developers to deploy application workloads on any cloud computing environment without worrying about what particular virtual machine environment that cloud service provider supports.
But once those issues are addressed, there remains a host of networking and database challenges.
Cisco has addressed some of the networking issues with the recent launch of the Cisco Nexus 1000V InterCloud, an implementation of virtual switching that makes it easier for a cloud service provider to extend the reach of its network fabric to the on-premise systems of its customers.
Meanwhile, TransLattice has created a database platform that can be federated across multiple instances of private and public cloud computing platforms. According to TransLattice CEO Frank Huerta, TransLattice Elastic Database 3.0 gives organizations a single instance of a database that can be globally distributed across multiple cloud computing environments.
Despite these advances, however, it will take several years before most IT organizations can upgrade their infrastructure environments to support dynamic hybrid cloud computing environments. In fact, most IT organizations are still debating what types of workloads should be run on private versus public cloud computing platforms.
“Right now we’re focusing our efforts on our own private cloud,” says Kirk Larson, CIO for Children’s Hospital Central California. “We’re still evaluating where public cloud service makes sense.”
Many application workloads that were initially developed on a public cloud are likely to migrate to either a private cloud or a dedicated managed hosting environment. As those transitions occur, IT organizations will find themselves managing multiple hybrid cloud computing scenarios at the same time.
“When it comes to cloud computing, it’s a mixed bag out there, especially when it comes to security. You have to do a lot of due diligence,” says Thomas Janus, CIO of PLS Financial Services. “The reason to put something in the cloud is to get it out sooner, not because it’s cheaper.”
The good news is that emerging cloud management frameworks, such as OpenStack and CloudStack, promise to simplify the management of hybrid cloud computing. The bad news is that the frameworks are still under development, so IT organizations will need to wait until vendors can roll-out mature cloud management applications based on the frameworks. It could be several years before any of these cloud management applications are truly enterprise class.
In the meantime, Accenture CTO Paul Daugherty says it’s critical for IT organizations to start preparing their IT houses now so they can support the hybrid cloud computing scenarios that will soon be available.
“Hybrid cloud computing is still an emerging technology,” says Daugherty. “But you need to start preparing now for a world where virtual machines and the applications that run on them have maximum flexibility.”