In the very first phase of WS2P connection when we start the service (startCrawling()), we try each peer one by one and most of the time without success. This delays a lot the start of the whole node's lifecycle (pull the network, ask for missing blocks, getting new peers, etc.). This is quite annoying.
Instead of connecting node by node, I suggest we try a bunch of nodes at once and then repeat the process. For example, we could try 10 peers in a row, then 10 other peers, etc.