Let us consider a cloud infrastructure shared by 1000 clients. Each instance in the cloud is dedicated to a given client. Let us assume a single instance can handle 10 concurrent requests. Also, the time window for processing a request is 5s, reflecting the required timeliness of a financial transaction or the shrinking attention span of a modern web user.
| Traditional | Erlang on Xen | ||
|---|---|---|---|
| Startup time of a new instance is 20s | Startup time of a new instance is 100ms | ||
| If more than 10 concurrent requests are received for a single client, only 10 of them get handled | Instances are provisioned after the request arrival – all requests get handled | ||
| At least 1000 instances must be always running (1 per client) and many requests are still left unhandled | No instances are running waiting for requests – the cloud footprint is zero | ||
| The size of infrastructure is dictated by the number of clients – at least 128 servers are needed | The size of infrastructure is proportional to the maximum load – 8 servers may be enough |