Scale Out Router
Scale-Out Architecture and NoviFlow Forwarding Planes
The old solutions just don’t work anymore…
Despite the last twenty years’ breakthroughs in data center compute and storage, or even perhaps because of them, rising network costs remain an inescapable issue. Networking technology lags significantly behind compute and storage in taking advantage of virtualization and disaggregation, resulting rigid all-in-one designs, vendor lock-in, the multiplication of protocols, proprietary interfaces, non-deterministic network behavior, and a very limited ability to interact meaningfully with the applications and services that use network resources. Unlike compute and storage, the network neither scales-up, nor scales out, placing limitations on the benefits of Hyperscale architectures and causing major headaches to data center and network managers
It’s time to disaggregate the router!
The NoviFlow Scale-Out Router Architecture breaks-up the monolithic router into three parts, each one optimized for cost and independently scalable according to changing network usage. The control plane (service cards) is replaced by economical off-the-shelf x86 compute servers. The backplane (chassis ) is replaced by low-cost standard L2 switches, and the expensive forwarding plane is replaced by off-the-shelf NoviFlow NoviSwitches, according to capacity and port density requirements. Note that all three offer the best price-performance profiles for the specific task at hand. Over a typical large scale router this results in savings of up to 75% on hardware costs and 65% in operational costs!
Why not run the forwarding plane on x86? The NPU-based programmable pipeline in NoviFlow switches is optimized for packet processing, executing many of the more complex tasks expected of switches and routers faster by a factor of as many as two or three orders of magnitude.
- Centralized MPLS/BGP routing application managing a distributed OpenFlow data plane
- Low risk, incremental, overlay deployment which could evolve to replacement of existing network equipment
- Supports millions of rules at the detailed flow level
- Supports up to 50,000 rule updates per second
- Packet processing done on NPU silicon not x86 software
- API to inject, manage and monitor routing actions
- Routing, tunneling, monitoring, security filtering, load balancing and other flow-handling runs at line-rate as part of network fabric
- Routing becomes an integral part of the network fabric
- Routing actions can be injected at any, or every, point in the network, reducing downtime due to human error and enabling easier centralized management of all routing in a network
- Lower CAPEX (up to 75% less) and OPEX (up to 65% less) over standard chassis-based solutions
- Genuine linear scalability of all three components of routing solutions (control logic, forwarding logic and data plane)
- NO vendor lock in as all components delivered on commercial off-the-shelf hardware.
Leave your email address or your telephone number below and we’ll contact you as soon as possible.