Limited resources
As we saw in the Identifying positive feedback loops section, resource constraints often form an essential step in exacerbating overload conditions. Some coding inefficiencies can be flushed out by realistically loading common operations, such as memory leaks or excessive disk usage. Others are more subtle and are best revealed by restricting resources.
Real-world example – 32 times 2 or 8 times 8?
One video conferencing company I worked for used standard servers that they owned and ran. I looked after procuring new blades, each of which ran with 64 GB of memory. We usually installed 8 lots of 8 GB memory sticks, but for one order, I bought two lots of 32 GB instead. That meant there were fewer components to go wrong, gave us more room for expansion in the future, and saved us some money.
The new blades came online and ran for several weeks before we got reports of higher packet loss rates. These were initially dismissed as poor networks (recall the...