When I talk to customers about the network challenges they are experiencing, they will inevitably start with how much more time it takes to access their files after the company has moved to a virtualized infrastructure or has implemented replication between data centers.
They follow that up by telling me that after reporting it to the IT department, the first attempt at fixing the latency problem is to increase the bandwidth of the wide area network (WAN). Inevitably, they find that doesn’t fix the problem.
While increasing bandwidth can improve network performance, it doesn’t address the latency that accompanies performance issues. That’s because data traversing the network comprises of many small packets of information, so making the pipe larger doesn’t help. The sender and receiver still have to wait for the packets to be sent back and forth.
Optimizing the WAN
In many situations, the answer is WAN optimization. It works by targeting three main issues:
- checks and acknowledgements
- targeting specific applications
- taking bits of traffic off of the wire.
Helps checks and acknowledgements: With TCP/IP, there are constant checks that occur to make sure the data is getting to where it is supposed to be. That tends to make the network very chatty, because there is underlying traffic always going on during data transmission. WAN optimization addresses the underlying traffic by fetching some of the packets before they are requested. That way, when they are requested, it will be local to the requester.
Targets specific applications: Most applications are very predictable (packet A is always followed by packet B, etc.), but they each work in their own way. A WAN optimization solution can use its knowledge of specific applications to perform similarly to TCP/IP — fetching predictable pieces of information to increase efficiency. WAN optimization also can identify unneeded redundancies within an application and send only the information that is needed at a particular time.
Takes bits of traffic off the wire: Say User A wants to access a Word document and User B, sitting in the same location as User A, wants to pull that same Word document. Rather than send that whole Word document a second time, the WAN optimization solution will understand that it has already has been pulled, and will only send the reference data across the wire. So instead of sending megabytes of information, you might just send a few kilobytes. It takes traffic off of the network, which reduces latency.
Wide Applicability: Just about any organization with multiple locations could be a fit for WAN optimization. For example, a company with offices overseas or on different coasts in the United States can experience significant network latency. I’ve also seen WAN optimization used by people in the fields of construction and engineering, which rely on sending very large files to job sites. Additional areas include law firms and healthcare.
Over time, we should see many more organizations looking to WAN optimization as a way to deal with the growth of desktop virtualization with more people telecommuting and using remote offices. With WAN optimization, organizations can ensure that employees can still perform their work in the same way as they would in the main office, without waiting for files and data.