The Growing Diversity of Virtual Network Operating Systems

SDN and network virtualization present more and more options to the forward-thinking enterprise.

By Arthur Cole | Posted Jun 24, 2016
Page of   |  Back to Page 1
Print ArticleEmail Article
  • Share on Facebook
  • Share on Twitter
  • Share on LinkedIn

The average enterprise may only have waded ankle-deep into the SDN waters to date, but already the diversity of options is making it difficult to determine the optimal solution for any given data environment.

Indeed, despite the underlying technology’s high degree of flexibility, organizations may wind up with multiple SDN frameworks that cater to the needs of individual workloads.

SDN functionality will vary according to the specific operating system that is used to manage the abstract architectures that are now sitting on top of raw hardware, and these solutions are as diverse as the vendors and open communities that develop them. Pica8 has one of the newest systems in the channel, version 2.7.1 of its PicOS platform, which is optimized for fast Ethernet switching over 25/50/100 GbE infrastructure. As such, it is best deployed on networks featuring either the Cavium XPliant or Broadcom Tomahawk switch, where it can drill down into data streams to enable highly granular flow management and optimization.

Meanwhile, both Cisco and Huawei are out with new network operating systems aimed at hyperscale environments. Cisco’s NX-OS is targeted at the Nexus 3000 and 9000 switches, providing a new Virtual Topology System (VTS) for streamlined overlay and network provisioning functions, as well as enhanced object store and model-driven API support. Huawei’s latest release, the Agile IoT solution, is targeted at IoT deployments and cloud-based SDN fabrics. It consists of multiple modules that coordinate gateway and controller functions and is available as an open, streamlined LiteOS version that provides rapid deployment and third-party development access.

Still, much of the open source activity surrounding the SDN OS is coming from the ONOS Project, which recently released the Goldeneye version of its software, which matches high performance and availability with multiple abstraction options. The system is designed to help service providers deal with diverse data sets coming from mobile, data center, cloud and other digital environments, in part by linking the multiple constructs that have arisen across both SDN and NFV deployments. These include the various traffic engineering and multicast content distribution solutions, as well as VPLS, SDN-IP peering and packet-optical convergence approaches.

But probably the most unconventional OS on the scene these days is the Apstra Operating System created by California’s Apstra Inc. The company actually bills it as the opposite of an SDN solution, because instead of re-engineering the switch to introduce new layers of abstraction, AOS works with existing switch architectures to enable SDN-like functionality through advanced automation and orchestration. In this way, the company says it can meet the networking needs of the digital enterprise without forcing network managers to learn an entirely new software-driven operational construct, while at the same time support multi-vendor environments without delving into highly specific configuration and integration issues. The system resides as a distributed platform that runs on top of legacy physical and virtual network infrastructure, providing end-to-end automation, workflow management and streaming functions by leveraging both turnkey and community code via RESTful APIs.

Ultimately, the growing diversity of virtual networking solutions can only help the enterprise as it transitions to new data-driven business models. Data and applications loads are becoming imbued with the ability to craft their own optimal working environments, so a one-size-fits-all SDN layer would soon find itself ill-equipped to deal with rapidly changing business environments.

But this also means the enterprise needs to think long and hard about what they hope to achieve with their networking infrastructure going forward, because decisions made over today’s implementation can have a big impact on tomorrow’s performance.

Comment and Contribute
(Maximum characters: 1200). You have
characters left.
Get the Latest Scoop with Enterprise Networking Planet Newsletter