Programmable switches for easier, faster and more cost-effective deployment of Network Virtualization
The inexorable rise of east-west traffic
The explosive growth in the use of NFV (Network Function Virtualization) and other forms of network virtualization for an ever-expanding array of applications has provided real benefits, but at the cost of increased network complexity, challenging long standing assumptions and ideas about how networks should be built.
A common SDN-based solution to mitigate this complexity is to deploy OvS (Open vSwitch) in servers, but that solution comes at an additional cost in terms of requiring the dedication of a number of server CPU cores to implement OvS, thereby reducing the number of CPU cores available for revenue generating activity. However, the exponentially increasing volumes of network traffic and increasing east-west traffic between servers are beginning to severely deplete server capacity and reducing the economies of virtualization.
NoviFlow vSwitch Offload solution leverages SDN and OpenFlow to implement a fully functional, centrally controlled and reliable network router that offloads OvS switching to a NoviFlow NoviSwitch at the ToR.
The NoviSwitch Difference
Unlike legacy network equipment, NoviFlow’s NoviSwitches have architected from the ground-up to deliver a networking fabric optimized for implementing programmable forwarding planes: very large table sizes, multiple tables, packet payload matching and traffic handling at line rate for even hyperscale networks.
NoviSwitches uses Network Processors (NPUs), which provide both high throughput and full programmability. When combined with NoviFlow’s NoviWare NOS, NoviSwitch provides a silicon based forwarding plane that is 2 to 3 orders of magnitude faster than x86 processors, yet can deliver the fully programmable packet pipeline needed to provide both the agility and scalability required by today’s networks.
This advantage is the basis for our vSwitch Offload solution. In this solution, NoviSwitches deployed at the top-of-rack transparently take on all packet processing and redirection tasks handling usually consuming one to several cores dedicated to running OvS in servers supporting NFV. NoviFlow’s vSwitch Offload connects transparently to OpenStack via the OpenFlow Controller using the Neutron plug-in. Consequently, vSwitch Offload can be deployed gradually, one server at a time or one rack at a time, over the entire NFV server base in a datacenter. This enables the data center to recover up to 60% of the capacity lost to OvS overhead, and to deliver the same processing capacity at less than 80% the cost!
- NoviFlow’s vSwitch off-load eliminates the server CPU overhead by moving the vSwitch processing from the servers to NoviFlow’s NPU based TOR switches.
- Also offloads ACL processing to the NoviSwitch, a major consumer of server resources in IoT, OTT and other client intensive services.
- Programmable pipeline, full OpenFlow 1.3 to 1.5, millions of flows at line-rate up to 100Gbps per port and 800 Gbps switching capacity in 1U
- Low risk, incremental, overlay deployment
- Routing, tunneling, monitoring, security filtering, metering, service chaining, load balancing, time stamping and other flow-handling at line-rate as part of network fabric
- Multiple port configurations available
- Server CPU performance gains from vSwitch off-load may be as much as 250% or a 60% reduction in the number of servers for a given payload
- Lower latency and less jitter: Port-to-port latency is lower and more consistent in a NPU-based TOR Switch than what can be achieved with vSwitch running on an x86 server
- Access control, path optimization and security become an integral part of the network fabric
- Simplifies deployment of NFV, cloud and multi-tenant services
- 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.