How do SD-WAN solutions improve network performance?

In the 2016 National WAN Report survey, respondents were asked to indicate the factor that most impacts their WAN is the WAN. Considering that unlike LANs, WANs have some performance-limiting characteristics, such as high levels of packet loss and latency, it is not surprising that two of the top five factors cited by respondents were performance-related. Since we are experiencing a fundamental shift from traditional WAN to SD-WAN, now is an important time to understand how SD-WAN solutions can improve network performance.

[[239428]]

Adding Bandwidth

Because of the way service providers charge for enterprise WAN links like MPLS, the typical WAN runs at megabit speeds, while the typical LAN runs at gigabit speeds. One benefit of SD-WAN is that it enables network organizations to close this gap by making full use of relatively cheap Internet bandwidth. There is no doubt that in some cases, increasing bandwidth can improve performance. However, the performance improvement is rarely linear with increased bandwidth, if at all. To understand what I mean, consider the example of sending a large file over a WAN link from Boston to San Francisco. If the size of the WAN link bandwidth is doubled, the time it takes to send the file may decrease, but it is unlikely to be cut in half. There are many reasons for this. One reason is the TCP window size, which is the amount of data that can be sent before the sending device stops waiting for an acknowledgment from the receiving device. Sticking with file transfers, it is likely that the file transfer will not be able to fully utilize the increased bandwidth and may only see a negligible improvement in performance due to the impact of the TCP window size.

Implement forward error correction and packet sequence correction

A factor that can have a significant impact on WAN throughput is packet loss. The impact of packet loss on TCP throughput has been extensively analyzed by Mathis et al. They provide a simple formula that oversees the maximum TCP throughput of a single session in the presence of packet loss. That formula is:

Where MSS is the maximum segment size; RTT is the round-trip time, and p is the packet loss rate.

The equation in the figure shows that as the round-trip time or packet loss rate increases, the throughput decreases. To illustrate the effect of packet loss, assume that the MSS is 1,420 bytes and the RTT is 100 ms. p is 0.01%. According to the formula, the maximum throughput is 1,420 kilobytes/second. However, if the loss increases to 0.1%, the maximum throughput drops 68% to 449 kilobytes/second.

One way to eliminate the negative impact of packet loss on WAN throughput is to implement an SD-WAN solution with packet-level forward error correction (FEC). Such a solution transmits a small number of parity packets. These parity packets can be used at the receiving end to reconstruct any lost packets, eliminating the limiting effect of packet loss on throughput.

Even if packets are not lost, but are delivered out of order to the receiving device, throughput will be reduced. The reason for this is that in most cases, TCP will force the sending device to resend the out-of-order packets. The way to eliminate this phenomenon is to implement an SD-WAN solution with Packet Order Correction (POC). POC dynamically reorders out-of-order packets on the receiving side of the WAN.

Leverage WAN optimization capabilities

In the related article, several use cases are mentioned where SD-WAN solutions with integrated WAN optimization capabilities can significantly improve application performance. One of these use cases is disaster recovery (DR). DR requires the transfer of large files between the primary and secondary data centers, which are separated by a large distance. Due to the impact of TCP window size on WAN throughput discussed previously, DR applications may not be able to transfer all the data required by the company's DR plan because they cannot fully utilize the WAN bandwidth. For many companies, a better solution is to implement WAN optimization capabilities with deduplication. Deduplication keeps the primary and secondary data centers synchronized, while the WAN link only needs to send the minimum amount of data.

Another use case is WAN optimization capabilities to support heavy protocols, such as Common Internet File System (CIFS). Heavy protocols require hundreds of round trips to complete a transaction. For example, let's say a business transaction requires 250 round trips. If that transaction takes 60 ms to complete over the WAN, the heavy nature of the transaction adds 15 seconds of latency to the round trip delay, and then the user will complain bitterly. Adding bandwidth increases costs but does not improve performance. WAN optimization capabilities can overcome the effects of heavy protocols and reduce the overall transaction time.

in conclusion

Network organizations can take comfort in knowing that any SD-WAN solution they implement will provide at least modest improvements in performance for some use cases. However, network organizations that are considering an SD-WAN solution and want more dramatic improvements in performance for a variety of use cases should carefully analyze potential solutions. They should focus on SD-WAN solutions that support advanced features such as FEC and POC, and that also offer integrated WAN optimization capabilities as needed, such as support for deduplication or reduction in overall transaction time.

Original link: https://blog.silver-peak.com/sd-wan-solution-improve-network-performance