So many technological changes are hitting the data center these days that it is becoming difficult to chart a path forward. Nowhere is this more evident than networking, which is rapidly transforming from a massive collection of switches, routers, cabling and software to a flattened, federated and largely virtualized construct.
But this leaves the average network professional in a bind.
While everyone wants to employ the latest and greatest technology, how is it possible to make this transition without bringing data operations to a halt? And what, exactly, will this new networking environment look like, let alone function and lend itself to regular maintenance and repair?
At the moment, the catchword for future networks is “composable.” While this may seem like more marketing hype from an industry that has long overpromised but under-delivered, there really is some meat to this concept. Under a composable scenario, users and even autonomous apps will be able to compose their own network architectures on the fly. MarketsandMarkets predicts this field will grow from about $616 million today to more than $5 billion by 2018, a 52.6 percent compound annual growth rate. To be truly composable, of course, the enterprise will need to virtualize the entire infrastructure stack, including compute and storage, which, again, will be difficult to do non-disruptively.
At the same time, we can expect a rise in the number of mini data centers as the Internet of Things (IoT) pushes data and applications to the network edge. These facilities will feature modular, hyperconverged infrastructure that will typically handle internal networking over the PCIe bus or some other interconnect. The challenge, however, will be to introduce LAN-like connectivity between these multiple centers and back to the central facility, which itself will likely feature high levels of modularity and hyperconvergence. To affect this change, network executives will likely rely on SD-WAN and highly targeted network services on the carrier level. Technavio estimates that the market for micro data centers will jump by 17 percent per year until at least 2021.
But if this is the way of the future, what are we to make of companies like Cisco that continue to roll out ever-larger core switches reminiscent of the old days of bulky network infrastructure? A look inside the chassis of systems like the Nexus 9000 provides the first clue. ENP’s Sean Michael Kerner notes that it can do far more than provide the large (400 GbE) bandwidth needs of a traditional data center. It also supports the company’s Application Centric Infrastructure (API) to craft highly specialized environments for today’s increasingly demanding applications. At the same time, the system incorporates machine learning and other forms of intelligent analytics to deliver automated, self-optimizing performance across a wide range of use cases.
Companies like Extreme Networks are also starting to stress agility over raw power in their data center solutions. The company’s new Agile Data Center platform offers a secure, plug-and-play framework that allows organizations to add new systems and capabilities at their own pace, all while maintaining automation, visibility and a host of other features across multiple domains and vendor products. The system features the Extreme Embedded Fabric Automation function that allows for the creation within seconds of any sized fabric that can then be customized for key workflows using Ansible and other tools.
Clearly, there is a lot to consider when it comes to the future of networking. For a while, at least, the upgrade path to the future will depend largely on the legacy systems that evolved in the past. Ultimately, however, all data networks should wind up in roughly the same place: extensible, scalable, highly virtualized and almost completely automated.
Arthur Cole is a freelance journalist with more than 25 years’ experience covering enterprise IT, telecommunications and other high-tech industries.