Virtualization, convergence and ‘miniaturization’ are three technology trends redefining the way networks are built and managed. As a result, re-shaping how services are created and offered.
Software-based networks and services, convergence of access network cores, and the flexible distribution of compute and storage resources each have powerful advantages when taken alone.
But when these technologies are combined, operators have an opportunity to create new service capabilities, realize astonishing cost advantages, and speed up service delivery by an order of magnitude. Who can afford to ignore such potential benefits, especially when competitors will be moving quickly to build an advantage with these tools?
Today I’ll introduce you to a proof-of-concept (PoC) system in which virtualization, convergence and miniaturization are applied to a representative network service. Integrated in the by WindRiver, Intraway, Casa Systems, Kontron AG and Aparna Systems, this PoC demonstrates that it’s possible to create an orchestrated, software-based, distributed and virtual CMTS that runs on high-density server platforms, while minimizing space, power and cooling requirements.
Rewriting traditional network functions and services as software applications
Intraway provides orchestration capabilities to deploy and manage the software constituting the virtual CCAP Core, which is the heart of a virtual, distributed CMTS system. The ability to automate instantiation, provisioning, monitoring and scaling of software services is essential to commercial deployments.
Casa Systems provided DOCSIS® 3.1 Remote PHY nodes and a software-based virtual CCAP Core. The virtual Core runs on generic servers in either an open source or vendor-supported cloud environment.
Operators that deploy their network functions as software-based applications, rather than hardware-based platforms, stand to gain compelling benefits:
- Updates, new features and new services can be created at the speed of software development instead of the speed of silicon development; days or weeks rather than months or years
- Network services and functions run on standard, commodity data-center style servers rather than expensive and proprietary purpose build appliances
- Software-based components can be distributed to any point in the network
- Scaling up, down or sideways is a matter of instantiating a new instance of an application rather than a truck roll or the deployment of a new box
- Operators have the option to create their own applications and services based on open source code, or to purchase applications off-the-shelf from virtual solution providers, and realize cost savings in either case
Certainly, there are business process, operational skill set and technical challenges to solve, but the potential benefits of virtualization are simply too great to ignore, and service providers and suppliers all around the world are moving quickly to adopt virtualization technologies.
Software running in the cloud
WindRiver developed Titanium Cloud as a complete product which delivers a cloud environment that’s stable, performance optimized, customizable, security hardened and above all solution-provider supported. Integration with a suitable hardware platform affords operators the choice to buy, rather than build, a ready-to-go solution in order to meet cost, time or expertise constraints.
OpenStack, on the other hand, is an example of an open source cloud platform that operators can use to develop solutions in-house. Such an approach enables infinite customization with no acquisition cost, although development and maintenance require investment and expertise.
Finally, below the cloud software layer, hardware is the ultimate terra firma of the network. Nothing runs on air, and a virtualized infrastructure doesn’t mean that there’s no hardware. A virtualized infrastructure means that generic, commodity hardware is used to run a cloud platform, which can then run every service and function that is built as a software application.
Miniaturization: Big power in small packages
Miniaturization is revolutionizing computing, once again. Chip densities continue to increase, enabling server manufacturers to offer unlimited choices in how compute, storage and network resources are packaged. More functions fit on smaller and smaller bits of silicon until entire computer systems fit on a single chip, with the size of that chip shrinking rapidly.
Yesterday’s super computer fits in a cell phone. Before long, an SD card will hold a terabyte of storage. Massive computing power can be delivered anywhere in the network from the headend all the way out to the customer, in any form factor and optimized for critical characteristics. Size, cooling, power draw, and compute resources can all be tailored to deliver an exact quantum of intelligence to any location in the network. And since hardware is standard, generic compute the resources can be scaled and replaced with ease, at low cost.
Such capability leads to infinite possibilities for smart cities, sensor networks, distributed data processing, and low latency services. Anything that can be imagined can be designed, built and deployed quickly and at low hardware cost.
Kontron’s SymKloud and Aparna’s µCloud 4015 are two example of small form factor, very high-density, scalable general-purpose compute platforms that are extending the frontiers of where network intelligence can reside. Such platforms will continue to evolve, limited only by the necessities of service provider imagination.
In this PoC the Kontron SymKloud was matched and integrated with WindRiver’s Titanium Cloud, while Aparna’s uCloud 4015 ran OPNFV OpenStack.
The interesting point here is that powerful and efficient hardware platforms are available that will support both buy and build solutions, reduce hardware footprint in headends, and enable very powerful distribution of compute across the entire network. Operators have the tools they need to develop new services and re-architect existing services at the speed of software on generic, commodity compute platforms.
Wrapping it all up
The speed of service creation is accelerating, the placement of resources is becoming infinitely flexible, and underlying hardware resources are becoming standard, homogenous, compact and far less expensive. Taken together, these trends add up to yet another revolution in the making.
For more information on the and this PoC, with your questions. As the Director of the Kyrio NFV Interop Lab at CableLabs, I’d love to chat more about how Kyrio can help you implement these technologies in your own infrastructures.
Make sure to register for a webinar on December 6 at 2 p.m. ET, presented jointly by Intraway, Casa Systems and Kyrio. I’ll walk through the PoC and provide tips on how you can transition your existing infrastructure to a more software-based one that reaps all the benefits that virtualization, convergence and miniaturization have to offer.