Servers capacity to handle a large number of concurrent connections can optimise a service resources footprint and mitigate the risk of denial of service attacks.

Apache Sling connection capacity is limited by its multi-threaded server model backed by kernel threads.

This talk will review approaches to increase the current limit and share practical experiments with a multi-threaded server model backed by fibers (lightweight threads) planned to be added to the JDK via project Loom.