Thank you to everyone who joined SDxCentral for its latest DemoFriday with HP on improving optimization with HP SDN applications. Attendees learned more about HP’s software-defined networking (SDN) strategy and learned how it provides its customers with a programmable network that is aligned to business applications. After the demo, HP was nice enough to take some questions from our attendees. Read the whole Q&A from the HP’s DemoFriday below and find out how HP SDN applications can enhance your organization.
What is the actual version of the HP VAN Controller?
HP: The current version of the HP VAN SDN Controller is 2.5.15. This is also the version that was used in the demos.
So is it supported with Cisco switches that support OpenFlow?
HP: Cisco switches currently don’t support both OpenFlow and normal forwarding simultaneously (Hybrid mode as described), so these applications wouldn’t be usable until that is fixed by Cisco.
How many apps do you currently have in the SDN APP Store?
HP: We have 20 applications, 18 partner apps, and two HP apps.
Is the VAN Controller already based on OpenDaylight or HP development?
Can HP VAN SDN Controller be managed by another service orchestrator like Planet Orchestrate for example?
Does the Network Protector app support controller redundancy (teaming)? I read in the release notes that controller teaming is not supported.
HP: The next release will enable teaming support on Protector.
Does the VAN Controller support multiple tables and Layer 3 routing?
HP: The VAN Controller supports devices with multiple tables and can program IP fields. The applications shown use normal mode for Layer 2 and Layer 3 forwarding, however.
Are there standards in SDN that each switch/node has to meet? Where would you suggest we find a summary of those standards?
HP: Most vendors support the OpenFlow 1.3 standard, but implementations vary, and only a subset of features are required. The Open Networking Foundation website is the source for specifications.
What is the scalability limits of this model? How many flows can be stored in the TCAM for each edge switch?
HP: The model only requires that a subset of the currently active flows to be supported in the TCAM, so the edge switch TCAM capacity is not a limiting factor. Normal mode is used for forwarding, so standard traffic patterns don’t consume TCAM resources. This is the beauty of hybrid mode approach to SDN, as opposed to pure OpenFlow architectures.
Do we need a specialized out-of-band management network from the SDN Controller to the switches?
What happens to encrypted traffic, i.e., in wireless environments with central WLAN controllers, when CAPWAP is in place? How does the controller handle this traffic at the edge?
HP: When the application provides the necessary dynamic control information, it doesn’t matter if the traffic is encrypted or not. This is why application-driven SDN is important, you can’t discover enough information by inspecting the traffic flowing through a switch.
Having legacy switches (without OpenFlow support), can we still use all the benefits of SDN?
HP: You would need to have OpenFlow to use SDN, however you can only deploy those at the edge for Network Protector and Optimizer where you’d need to get the benefits.