Network Load Balance with OpenFlow
May 21, 2011
James Hamilton had an post on Network Load Balancing with OpenFlow. In this post, James re-iterates the vision of migrating networking to open and commoditized platforms, which we not only share but work hard to make it happen.
Open, multi-layer hardware and software stacks encourage innovation and rapidly drive down costs. The server world is clear evidence of what is possible when such an ecosystem emerges.
This post also gives an interesting example of using OpenFlow to load balance a multi-campus network.
… Essentially they distribute the load balancing functionality throughout the network. What’s unusual here is that the ideas could be tested and tried over a 9 campus, North American wide network with only 500 lines of code. With conventional network protocol stacks, this research work would have been impossible in that vendors don’t open up protocol stacks. And, even if they did, it would have been complex and very time consuming.
I find an interesting point in James’s conclusion – if it is not OpenFlow, it will be difficult (complex and time consuming) to try out the load balancing idea.
The question about the value of OpenFlow was often raised as an attack point to this new technology. People often quoted Professor Scott Shenker, a founding member of ONF, “[OpenFlow] doesn’t let you do anything you couldn’t do on a network before”. Some said OpenFlow was nothing new and did not bring value to the network world. This is only half of the truth. Take the load balance case as an example, that feature can be done on selected Cisco switches if you hire a network specialist for a couple of weeks, if not months, to try out various switches and configuration.
With the example of implementing a load balancing network with less than 500 lines of code, we see the potential of this OpenFlow protocol.
- New features can be done and tune through software, without involving manual configuration. This gives the hope a scalable network can be self managed and tuned by software.
- By allowing researchers and engineers to try out the new ideas on a production network, OpenFlow can roll out new network innovation much faster than traditional network environment.
This implies someday the system administrators might be able to use software, through OpenFlow protocol, to replace some tasks that can only be carried out by network specialists. The system administrator can focus on more critical things like optimizing the efficiency of the whole data center or planning for the future growth than trying to trace down the packets on the wire.