IBM has announced the availability of the Programmable Network Controller, a new OpenFlow-based product for software defined networks (SDNs). The software-based controller supports device discovery, multitenancy and fail-over capabilities. IBM already offers OpenFlow-enabled switches, but this is the company's first IBM-branded controller. At nearly $100,000 for a license, IBM seems to be pricing the product for service providers and large enterprises.
The Programmable Network Controller supports OpenFlow version 1. 0. The company says it can manage up to 100 OpenFlow switches, including those from other vendors, as long as the switches support the 1.0 version of OpenFlow. This means IBM's controller should work with network devices from Cisco, HP and other vendors.
The controller is offered in a dual active/passive fail-over configuration, which the company recommends for production environments. Most chassis switches, including Cisco's Catalyst 6500, are deployed with dual supervisor modules for redundancy, and an OpenFlow deployment is no different. The controller also supports multitenancy so that service and cloud providers can ensure separation of customer management data and customer traffic. Enterprises can also use multitenancy to delegate management to distributed IT departments.
In January 2012, IBM and NEC announced a partnership to deliver an OpenFlow product set using NEC's ProgrammableFlow Controller and IBM's G8264 top-of-rack switch. IBM won't say whether its controller is based on NEC's product, but given the similar specifications and functions, and the two companies' close working relationship, it's likely that Programmable Network Controller is from NEC. NEC's ProgrammableFlow Controller is one of the few OpenFlow controllers on the market, and the company has been involved with OpenFlow bake-offs and interoperability testing. The NEC controller won the Best of Interop 2012 award this May.
Using OpenFlow promises to make networking more effective at interconnecting applications and maximizing network capacity by taking the shortest path between two points. Traditional networks use up and down links to move traffic to hosts between switches. The north-south traffic means you need increasingly more capacity between switches to avoid choke points. By interconnecting edge switches, IT can increase east-west traffic and improve performance by taking a more direct path between hosts. OpenFlow also provides a basis for defining paths through the network based on business rules. In a simple example, a company might want to ensure that videoconferencing traffic always takes the fastest route, while email can tolerate some delay. With OpenFlow, traffic paths can be redefined to meet performance needs. OpenFlow also provides the basis for developing these rules via software, which in turn enables the dynamic creation, movement and deletion of virtual machines with minimal impact to network configuration.
Up Next: What's Missing
