At ARCHIMEDIUS I’ve talked about the challenges facing Cisco over the next five years, especially the rise of software-defined networking (or what some of us have been calling infrastructure 2.0). As networks become more elastic or dynamic, custom software becomes more important than custom hardware.
Think back to the glory days of SGI’s multi-billion specialized graphics workstation-driven market cap before the rise of Wintel and today’s sub $300 million market valuation. Now fast forward to the likes of Cisco, Nicira (recently acquired by VMware) and other startups including PLUMgrid and Big Switch.
We discussed the implications in Golden Age or Golden Fleece? back in April.
Besides the nimbler world of software-centric networking (and how dynamic markets tend to favor more innovative startups), the core hardware-centric network is now competing directly with server-centric and services-centric players (including IBM and HP). This is more than a significant development, it is a harbinger.
Archimedius- April 2012
With the billion dollar acquisition of Nicira weeks after this post, things are coming into even sharper focus. Yet if you think the rise of virtualization has transformed alliances into rivalries; take a look at the potential of hybrid cloud. Hybrid cloud is an operating model of cloud computing whereby IT teams run multi-tier apps seamlessly across data centers and public clouds as if they were a single, highly elastic data center. For now let’s call a hybrid cloud a boundless data center.
The Boundless Data Center is a Hybrid Cloud
Imagine an organization no longer preoccupied with spending the capital (and time) to plan and build a physical data center, a process that takes more than 3 years on average and can cost tens of millions of dollars per megawatt. Imagine instead that it can use public clouds (yes, multiple) tactically and strategically to manage demand spikes and organic growth. As workloads are optimized across more facilities (based on everything from changes in power costs, to access to users and/or imminent storm fronts to even mundane items like changes in cloud service provider terms of service) IT assets are used more efficiently.
Organizations do not have to over plan or overprovision for periods of occasional high use. Remember “Own the base, rent the spike” from the Future in Review panel on What Every CIO Should Know about Cloud? The network is strategic to this vision, perhaps more strategic than ever. Yet companies like VMware are changing the rules for the network in the same way that they have for other hardware-centric models, including servers. That means that software and applications are poised to have considerable influence over the direction of hybrid cloud, along with software-defined networks, or what we’ve called infrastructure 2.0.
The transition of a legacy multi-tier application from being orphaned in a network to being run in a hybrid cloud is mostly about software and services, not hardware. These new demands make elastic, software-defined networks (and their services) even more strategic to the evolution from the Public versus Private Cloud Trap to the robust, secure and elastic hybrid cloud.
We are not there yet. Much work remains to be done, most notably the evolution of software that can clone app clusters and services into and between data centers and private and public clouds. This capability would be one of the first steps in the coming hybrid cloud revolution.
For Cisco and the other hardware-centric networking players, this new boundless data center era promises to be as disruptive as virtualization was to the server industry and the personal computer was to the mainframe industry. The question is: will the boundless data center help today’s hardware-enabled leaders to maintain their leadership or will it set the stage for yet another software-enabled revolution?