TCP Protocol Optimization and WAN Optimization
by, 12-11-2009 at 03:42 PM (8057 Views)
In addition to some of the other technologies already discussed in other blog articles here, such as Byte Caching, Object Caching and Compression, TCP Protocol Optimization is another key technology used by WAN Optimization Appliances.
What is Protocol Optimization?
Protocol Optimization is the use of in-depth protocol knowledge to accelerate user response time. By understanding the intricacies of how specific protocols function, WAN Optimization appliances can anticipate
user requests, which results in data being retrieved before clients request it. For applications in which requests are typically serialized (HTTP, CIFS, data backup and restore), or for traditionally “chatty” applications originally designed for
LANs (CIFS, MAPI), the performance improvement gained by Protocol Optimization can be considerable.
Why should I perform Protocol Optimization in my network?
In today’s enterprise, latency has become a bottleneck to user productivity. Whether delays are due to remote Web based applications or high-latency protocols such as MAPI and CIFS that typically experience degradation when accessed across a WAN, Protocol Optimization can greatly improve performance. When discussing latency, it
is important to understand that while bandwidth can sometimes be a factor (if your WAN is constrained), latency is more commonly caused by the sheer number of roundtrips required to satisfy a request.
All the WAN Optimization techniques together can effectively minimize delays associated with waiting for data retrieval by anticipating user requests. The result is that when clients make a request, the data can be served at LAN speed instead of WAN speed. This allows users to stop waiting for data retrieval and start being more productive.
How does Protocol Optimization work?
Protocol Optimization is possible because SG appliances have the unique ability to terminate user requests. Many vendors simply repeat the exact same request made by the client, with no real intelligence
about the how the protocol actually works. Blue Coat's ProxySG appliance, however, terminates the request from the client as if it was the server and opens a separate connection to the server that allows it to intelligently make requests on behalf of the client. As a result, ProxySG appliances apply Protocol Optimization to many of the protocols that they terminate. This includes HTTP, HTTPS, CIFS, MAPI, FTP, MMS, and RTSP.
CIFS Optimization – Read Ahead and Write Back
For enterprises accelerating CIFS, Blue Coat’s WAN Optimization technology has the ability to reduce user response time by implementing Read Ahead and Write Back. The biggest problem generally associated with CIFS is that the Windows client read is often limited to 4KB. While requesting a file with 4KB reads over the LAN may not be
noticeable, accessing the same file over the WAN can be painful. To minimize the effects of this client limitation, ProxySG appliances take advantage of the ability of the CIFS protocol to support 60KB reads. When a client read request is received, the appliance anticipates future reads and makes a 60KB bulk request, or Read Ahead, thus saving
numerous round trips to the server. Write Back is the method used to minimize response time when a user is writing (saving) a file to a server. Most often, the latency associated with writing a file to a server is due to the acknowledgements required from the server in order for the client to continue sending additional data. With Write
Back, the ProxySG appliance acknowledges the data as if it was the server, decoupling the communications between client and server, allowing the client to write the file with little to no latency.
Consider a client attempting to request a 20MB file over the WAN. If the CIFS client read is limited to 4KB, the client will require over 5000 reads to retrieve the entire file. Over the LAN, the delay per round-trip will not be a significant portion of the transfer time, but requesting this same file over a WAN with 300ms latency will take 25 minutes! With a ProxySG appliance deployed, Protocol Optimization is able to retrieve this file quickly by anticipating the many
requests that would be necessary to retrieve the entire file. Instead of making hundreds or thousands of requests as the client is required to do, the SG appliance is able to read the data in large chunks instead. The result is that the ProxySG appliance is able to retrieve this file over the same WAN in less than 2 minutes.
The Blue Coat Difference
Because Blue Coat ProxySG appliances are application proxies, they have the ability to maintain application level persistent connections on both the client and server-side connection. This results in less time spent creating and tearing down connections, effectively reducing latency even further.