Hybrid Clouds and the Networks that Love Them
Many call the hybrid cloud the next big thing in IT development. But while the popular impression is that this proceeds from the natural convergence of public and private resources, the fact is that hybrids will require a fair amount of infrastructure development of their own.
One of the key elements, of course, is networking. Software defined architectures, or so we’re told, will be able to pull network resources from wherever they reside, but that technology is still several years off for much of the enterprise. The need to build hybrid clouds, on the other hand, is front and center.
According to Infonetics Research, hybrid cloud adoption is set to double over the next year as organizations seek the agility, scalability and cost efficiencies needed to cope with Big Data, mobile traffic, and other data-fueling activities. The big question, though, is whether top network vendors are ready to step up with solutions that can foster the kind of network/server/storage interoperability and orchestration that hybrid environments require, preferably without a full rip-and-replace of existing infrastructure.
From the looks of things, the networking community has only just come to appreciate the needs of the hybrid cloud. Late last month, Cisco introduced the InterCloud software stack, designed to enhance workflow migration and sharing across the cloud. At the same time, its Application Policy Infrastructure Controller seeks to unify disparate application management systems so they can more easily coordinate workflows across in-house and third-party infrastructure. Both systems are likely to emerge as key components of Cisco's Application Centric Infrastructure (ACI), which seeks to foster broad infrastructure integration across disparate data environments.
In an odd twist, Cisco’s push into hybrid cloud architectures is not so much furthering its competition with other networking companies as it is drawing new battle lines with hybrid platform providers, particularly those pursuing solutions based entirely on open standards. A key rival, then, is Red Hat, which recently unveiled the Cloud Infrastructure 4.0 and Enterprise Virtualization 3.3 releases as part of the Linux OpenStack Platform 4.0 system. Red Hat strives to build its hybrid cloud environments completely on commodity hardware, which would include the networking infrastructure. To that end, the company has been devising open systems for four key areas of the hybrid cloud: traditional virtualization, hypervisor management, IaaS and converged cloud infrastructure.
This friction over the underpinnings of hybrid architectures is also a key driver of the increasingly large wedge that exists between Cisco and longtime frenemy VMware. VMware’s vCloud Hybrid platform utilizes the same hypervisor-centric approach that the company is fostering with its SDN portfolio, which means it prefers its hardware to be as basic and uniform as possible: capable of pushing packets, supporting multiple hypervisors, and not much else. In that vein, VMware is more aligned with commodity hardware suppliers like Intel, IBM and the rash of vendors building new ARM-based modular components, provided they maintain strict adherence to industry platforms that VMware supports, namely OpenFlow and OpenStack.
Networking in the hybrid cloud is likely to be a very different animal than in traditional data center infrastructure. Not only will it need to move data from place to place more quickly and efficiently, but it will have to contend with a swirling mix of virtual constructs and policy/governance architectures stretched across wildly disparate multi-vendor platforms.
Ironic, isn’t it, that even though much of the functionality of the data environment is moving up the stack, it still relies on the smooth interaction between software and underlying hardware?
Photo courtesy of Shutterstock.