WAN symptoms in question are latency, packet loss, disconnects, bit errors, etc. usually exist on Wide Area Networks and as a result cause applications performance degradation. But our WAN’s essentially are LAN’s connected with each other and those connections impose geographical distance between those LANs, thus creating “WAN behavior”. Inside individual LANs those issues happen rarely and if they happen – we have teams that are ready to go and trouble shoot them right away. The issues may be caused by malfunctioning hardware or faulty software running on the hardware.
When do these issues become as important in LANs as they are in WANs? Probably when our LAN becomes huge – such as Cloud computing Data Centers.
These Data Centers are interconnected with each other and one of the main differences with our traditional Data Centers is that they are managed automatically by very complex and smart management systems. These systems are capable of automatically allocate hardware and software resources for our applications, maintain these resources and expand/decrease capacity of the infrastructure based on the application demands almost in real time.
So imagine that the system is scheduled to perform an OS upgrade on the managed infrastructure: it will download OS images on thousands of machines thus creating huge traffic congestion within the Data Center. What will happen to the system that knows only how to work on LAN and was never tested on coping with delays? Packet Loss? – it might go down partially or completely. Since it was not tested – how can we know? (Sounds a little like Skynet scenario in Terminator, doesn’t it?)
The importance of the management system performance becomes utmost in this case and it makes sense to test how the system will react to the issues that now might happen within it, doesn’t it?
This is where WAN emulation solution can help – it can create those conditions within the Data Center or between the Cloud Data Centers and the “unexpected” can become “expected” and fixed.
Leave a Reply