No matter how carefully you design and build your network, there is always a chance that packetloss will occur somewhere. Why does packetloss occur? Well, the most common causes are:
- Congestion – simply put, the network is overloaded and network elements are dropping packets that they cannot process in time. This is typically first seen at peak hours or during OS updates.
- Microbursts – the same theory as congestion, but is caused by very short bursts of activity that might be difficult to see without suitable tools. The microburst will overload network elements for a few milliseconds, but this is enough to cause packetloss and bad subsequent consequences.
- Coverage – if the final hop is via a wireless link, 3G/4G/5G or wifi, then there is the possibility of interference, loss of coverage or handover between access points that all have the possibility to introduce packetloss.
The reality is, much of what happens on a network is out of the control of network operators. Sure, the physical links, choice of MPLS, VLAN and perhaps internet BGP routing all are, but beyond that, its outside an organisations control, since most services, such as TCP are an end to end protocol.
If your network is suffering from disappointing performance and end users are seeing packetloss, then help is at hand. By taking control of how TCP handles packetloss in your network, you can dramatically improve performance. TCP Acceleration also manages congestion by optimising TCP throughput. The overall impact is a faster, more responsive and a more consistent user experience.