State of OpenFlow 2012 – Testing and Analysis

What is the state of OpenFlow as we transition into 2013?

Router Analysis, Inc. will be taking the last few weeks of 2012 to put together a comprehensive report on the state of OpenFlow.  Our goal is to work with as many analysts and academics to understand where OpenFlow was expected to be vs where OpenFlow is.  While we will do OpenFlow testing, if anyone is expecting a bake-off type of report they will be disappointed, OpenFlow switches are just not at the performance level that a bake-off makes sense.

The skeleton of the test plan

  • Setup and Basic Configuration
  • Enabling OpenFlow with the BSN Floodlight Controller
  • Enabling OpenFlow with RouteFlow from CPqD
  • Commands and outputs
    • show the switch OpenFlow configuration
    • show what a flow looks like once programmed into the switch
    • show any error messages that come from switches when incorrect/redundant flows are added
  • Test the flow table limits of the switches based on the statements from vendors.
  • Test the speed of programming the same 1k flows into each switch
  • Program all switches at the same time through Floodlight if possible
  • Run the switches in OpenFlow only mode and Hybrid Mode if possible

Since we are going to base the main tests on the lowest common denominator we are expecting to program a maximum of 1k flows for any cross-vendor tests.

What we expect to see

Our expectations are set in the middle, we have two vendors who publicly support OpenFlow today and one vendor who has it in academic/beta use.  We expect that each vendor will meet their stated number of flows and that performance will vary quite a bit due to the differences in hardware.  As we are completely community driven and bootstrapped, we cannot afford to buy the latest/best version of hardware from each of the vendors, so we are relying on either the vendors to provide the equipment or generous donations/loans of equipment for the test.

Our Current Hardware Lineup

Pica8 Pronto 3290 – This is an older switch, but a very solid performer and well tested.

HP 6200yl – This is also an older switch but shows as “updated” on their website and supports the current OpenFlow enabled code from HP

Unknown Brocade Switch – We are still waiting to hear what Brocade will be able to provide.

We are accepting equipment from any other vendor with a public plan to support OpenFlow i.e. Juniper, Cisco, Huawei.

We are also open to support from any test vendor, we can run all of the tests without one, but having a test vendor involved would make things a bit easier.

Summary

We are going to test OpenFlow as it stands today.  We are going to use Floodlight and Nox.  We are going to test Pica8, HP, Brocade and anyone else who wants to join in. We are going to run traffic, we are going to put in flows, this will be real life captured flows installed into the system using our in-house Intelligent Flow Tool.

We are working with analysts and journalists to determine what else to add to our report.

The final results will be out at the end of the year, early January.

Updates

Pronto and HP using Floodlight and the Router Analysis Flow Test Tool

State of Openflow 2012 Notes Day 1/2

RouteFlow

State of Openflow 2012 Notes Day 3

More RouteFlow

State of Openflow 2012 Notes Day 4

HP and RouteFlow

State of Openflow 2012 Notes Day 5/6

Trackbacks

  1. […] IWNetworks has provided Router Analysis with an SDN 8952S switch.  The SDN 8952S is a 48 port 10GE + 4 port 40GE line-rate capable switch with OpenFlow v1.0 support.  In our testing we were able to connect the switch to our RouteFlow setup and our Floodlight setup without any issues.  As this is not a speed test, we did not do any benchmarking of the IWNetworks SDN 8952S.  We are happy to have IWNetworks join in the State of OpenFlow 2012 Testing and Analysis. […]

Leave a Reply

Your email address will not be published. Required fields are marked *