Global network traffic is exploding faster than complaints about Pokémon Go availability. Cisco’s Visual Networking Index (VNI) predicts global IT traffic will pass a zettabyte ([ZB] 1000 exabytes) in 2016, and will reach 2.3 ZB by 2020. The same report mentions that smartphone traffic is expected to exceed PC traffic by 2020. That was before Pokémon Go; we’ll have to wait and see how its rapid rise in users will affect that prediction.
Regardless, no one disputes that traffic is on the rise. It’s a given with an increase in cloud usage by businesses and consumers alike, as well as the steady arrival of new and interesting IoT devices (Ericsson Mobility Report 2016 predicts a rise from 5.3B to 16B devices by 2021) that traffic generation and delivery is going to continue to grow.
Networks, particularly carrier networks who are most impacted by this growth, need to scale and scale quickly. That’s a requirement F5 hears from customers every day across every vertical industry and from all sizes of organizations. It’s not just bandwidth, which is important, but it’s the scale of the many services offered by carriers and enterprises that will need to scale to support more users, more traffic, and, invariably, more attacks. That means services like load balancing apps and infrastructure, security scans and scrubs, and generally quality of service enforces will need to scale.
But this scale is not linear nor is it necessarily evenly distributed. It is often sudden and unpredictable, and puts varying degrees of load on different parts of the infrastructure. Cisco’s aforementioned VNI report, for example, notes that “Busy-hour Internet traffic is growing more rapidly (51% increase) than average Internet traffic (29% growth).”
That means scale of all the services between users (that includes things and systems as well as consumers) must be adaptable, able to meet demand in a dynamic way while not commensurately increasing costs. That means some form of automation and orchestration to enable seamless scale of services when they’re needed the most.
For carriers and some enterprises, that means network functions virtualization (NFV). NFV is based on the premise of virtualized network services that can be launched, scaled, managed, and decommissioned on-demand using software-defined methods. Programmability is key, of both the systems governing the scale as well as the network appliances providing security and scaling services.
But speed is still one of any network operator’s prime directive. No carrier wants to be faced with its own version of a network function Kobayashi Maru, a no-win situation in which they can only choose speed or security, scale or software.
Advances in hardware like Intel® Communications Chipset 8955 (formerly Coleto Creek) are critical to enabling both the scale and security network operators need to ensure fast, reliable, and secure transport of its growing load of traffic. Whether we’re deploying those services as virtual instances on our carrier-grade VIPRION hardware or as virtual appliances using orchestration frameworks like OpenStack, F5 relies on Intel hardware to improve the speed with which critical services can be applied. Designed to accelerate mathematically complex functions such as specific cryptographic calculations and compression algorithms, Intel® Communications Chipset 8955 ensures that application delivery controller (ADC) functions like load balancing and security continue to operate at scale and speed. By taking advantage of such hardware, even when an F5 BIG-IP is deployed as a virtual appliance, it can continue providing fast security and scaling services for both applications and critical infrastructure services, like domain name system (DNS).
Whether the network is carrier or enterprise grade, speed and seamless scalability are both desirable to handle the growth in traffic. While there are cost advantages to leveraging NFV technology and its programmatic foundations to seamlessly scale services on-demand, it remains true that hardware, whether custom or commercial off-the-shelf (COTS), continues to benefit from advances by Intel to provide the much-needed acceleration of the complex, compute-heavy functions that keep the network reliable and secure.