Oct 17, 2011
I never look forward to leaving at the end of a ball game. Not because I don’t want the game to end, but because I dread the mass exodus of tens-of-thousands of people congesting aisles, corridors and parking lots.
This “bottlenecking” affects the capacity of the entire stadium and is typically limited by the number of exits and appropriate directional resources. The term “bottleneck” is taken from the “assets are water’” metaphor. As water is poured out of a bottle, the rate of outflow gets limited by the width of the channel of exit—in essence, a bottleneck. Increase the width of the bottleneck and you can increase the rate at which the water flows out. As this relates to a data network, a network connection is “bottlenecked” when data flows fully consume or saturate the link.
Similar to the sports stadium analogy above, bottleneck (congestion) has the same adverse impact on WANs. With the ever-increasing growth of applications, users, and data volume, networks are pressured to support the growing traffic quickly and reliably. This traffic is creating network bottlenecks that lower business productivity and restrict access to applications and data.
Network congestion occurs when a network link carries excessive packets that deteriorate the quality-of-service. Some of the resulting problems include queuing delays, lost and out-of-order packets, and the prevention of new connections. These problems also lead to incremental increases in bandwidth overhead, which inhibit network throughput.
Some transport protocols–like TCP–use re-transmissions to counter packet loss, but they continue to add congestion to the network even after the initial load has been reduced. Why? Because when packet loss occurs, end-points send additional packets that repeat the lost information, thereby increasing the data rate sent. When you are in the middle of dealing with network congestion, sending additional packets is the last thing you want.
Advanced WAN optimization solutions solve network congestion problems to improve application performance, increase deployment flexibility, and enable business agility. To overcome the adverse effects of network bottlenecks, WANop appliances use various optimization technologies, including quality-of-service (QoS), traffic shaping, deduplication, and real-time error correction.
QoS and traffic shaping are the oldest techniques of the ones listed above. These help avoid the negative effects of network congestion by applying priority rules to give certain applications higher priority than others–and to ensure specific traffic is allocated to appropriate network resources
Deduplication, like Silver Peak’s patented Network Memory technology, is newer. Network Memory inspects all traffic sent between clients and servers, storing the information as a local instance in the WANop appliances. Repetitive information is delivered locally rather than sent across the WAN—improving application performance and WAN utilization. Bottleneck averted!
Silver Peak is unique in that it also provides real-time techniques to address packet delivery issues common to shared WAN technologies, such as MPLS and IP VPN. These include adaptive Forward Error Correction (FEC) and Packet Order
Correction (POC) to overcome dropped and out-of-order packets. These techniques rebuild packets in real time instead of requiring re-transmissions over the WAN, which has a substantial impact on networks with high congestion. Because they operate at the IP layer, these techniques address all IP traffic—another advantage over aggressive TCP re-transmission techniques.
Finally, almost all WAN optimization solutions mitigate the impact of latency across the WAN by using various well-defined TCP acceleration techniques, like adjustable window-sizing and selective acknowledgements, as well as CIFS acceleration techniques. These technologies overcome inherent chattiness that can hamper application performance across a WAN.
If you are responsible for your network’s performance and reliable delivery of applications and content, remember, WAN optimization solutions have the “traffic management” technologies for reducing the adverse impact from congested networks.