Summon servers to meet demand

Many websites will use one big server for everything. Why pay for something that's only going to be fully utilized during peak hours? With our model, we spin up several small servers when traffic increases, then shut them down when things settle. Not only is this cost effictive, it's safer. One server out of 10 crashing is much less worrying than your only server crashing.

Lets look at a real example

Graph showing CPU spike

Here we see in increase in CPU usage on the two running web servers.

Graph showing addition servers booting up

Before the servers max out at 100%, additional servers are brought online to handle demand.

Graph showing CPU normalize

Traffic normalizes as we find a comfortable amount of computing power to handle the burst in traffic.

Graph showing servers shut down

The traffic surge is over, terminate the idle servers.

Graph showing normal operation

Return to normal operation, around 30% CPU usage.

All of this occured in less than one hour. The site went from two servers, to 10 servers, back down to three servers. All without human interaction or interupted service.

The client was billed for minutes of server usage, not hours, days or months.