Unlocking the Communications Industry with NFV and Pure-Play OpenStack
Telcos and service providers need this because the ground is shifting under them. While many operators have embraced virtualization as a tool for increasing operational agility, first-generation approaches to network application virtualization — dependent on expensive proprietary hardware, complex and inflexible application stacks and tools — are aging badly. Like prior generations of telecom infrastructure, they tend to lock operators in, slow them down, and keep costs high, even as upstart competitors (the WhatsApps of the world) use new tools to build greenfield products that intrude on CSP turf (e.g., by disrupting SMS) and piggyback on CSP-promoted innovations (like 4G on smartphones).
But telcos now have a clear path forward, driven by many parallel technological advancements and industry dynamics:
- Hardware innovation beyond the Moore’s Law limit: catapulting forward the performance and accessible sophistication of standard compute and network platforms, the power of development tools and standard libraries (such as Intel’s Data Plane Development Kit) to enable robust virtualization of network functions in cloud-hosted software.
- The exploding phenomenon of OpenStack, both its own project and member ecosystem and closely-linked ecosystems surrounding Linux, open hypervisors, vSwitch, SDN and related projects, like OpenDaylight and OPNFV (which Mirantis has sponsored since 2014).
- Among solution providers, increasing understanding of cloud-native application engineering techniques and the possibility and need to create a whole new generation of cloud-native VNFs and orchestration tooling.
Not a Mirantis PivotIn prior years, an announcement that a software company like Mirantis was getting into the carrier space might have heralded a big-deal pivot. We might be talking about forking OpenStack; moving to purpose-built, fault-resilient hardware and proprietary silicon. We might be talking about changing our business model: developing more complex partnerships; committing to hard-certify partner products under simulated loads; building out new services, support, sales and training organizations.
Today, a pivot of sorts is certainly happening — but it’s happening in the carrier space. At Mirantis, we’re actually talking about leveraging our pure-play OpenStack business model; years of OpenStack code-contribution and project leadership; more years of OpenStack cloud-building experience (including work with some of the biggest global carriers and their largest suppliers), all boiled down into the Mirantis OpenStack distribution — plus the model of our expanding Unlocked partner program: all working together to meet the requirements of the NFV use-case.
Given access to all this stuff (and all this knowledge) Mirantis and its partners can now deliver a narrow, but expanding set of validated OpenStack-based NFV solutions that meet carrier requirements for performance and reliability — simply by deploying Mirantis OpenStack and partner NFV/VNF solutions in (admittedly-clever) reference configurations on commodity hardware. We can do it because these solutions are using the same techniques (though in some cases, a slightly different kit of parts) as OpenStack solutions for large-scale web app deployment, enterprise IT, big data, etc. And we can do it because carriers are also changing: embracing problem-solving approaches and technology architectures that work well with these new tools, though they work very differently from artefacts of the carrier legacy plant.
To illustrate the point, in an interview last week with Mirantis’ John Jainschigg, Martin Taylor, CTO of Metaswitch, went quite deeply into the cloud-native mechanics of Perimeta SBC, articulating quite clearly the advantages of using cloud-native engineering approaches to build and scale very-high-capacity, function-rich VNFs (Perimeta is a vSBC — a virtualized Session Border Controller). It’s definitely worth a watch.
Metaswitch, whose validation of Perimeta is part of our NFV announcement, is a good example of how our VNF Partner Ecosystem brings carrier-market-specific value to the table. They have deep experience, dating back through several generations of post-conventional telco solution strategy, and they know a lot about how to build VNFs in software (both for command/control and media processing) and run them on VMs. In a larger sense, though, our NFV partners are partaking of the same Mirantis Unlocked partner value proposition that works for IT, dev/test and other use-cases. Well-defined, self-executed, Mirantis-reviewed validation process; well-defined integration pathways, standard deliverables; and methods for automating deployment (e.g., via Murano, Fuel Plugins, etc.) work for carriers as well as they do for other Mirantis customers. The Mirantis Unlocked Partner Program, which Metaswitch, Overture and Citrix joined as part of today’s news, lets lots of solutions play, while ensuring quality and smoothing joint support. They promote customer choice. And they let us focus on Mirantis OpenStack.
Mirantis OpenStack itself is, of course, the fulcrum for our NFV work. Our current Open NFV Reference Platform draws on our earlier focus on deploying OpenStack with HA, minimizing cluster footprints, adapting to NFV-friendly network-hardware-level innovations like SR-IOV for NIC port virtualization and OS/hypervisor offload to hardware (also very useful for storage networking). In projects live with our carrier customers today, we’re successfully enabling VNF access to advanced CPU features (e.g., NUMA, CPU pinning) and tools like Intel DPDK. Over time, these field-proven evolutions will work their way back into Mirantis OpenStack, and to our upstream submissions.
Taken together, we think the Mirantis NFV initiative provides the right components to help CSPs get past their legacy, innovate and offer new services faster, and consolidate operational efficiencies that will shore up their bottom lines.