Cache update function is not parallelized and could lead to contention when there are a lot of peers in cache.

[mwyzhao]

As the peer cache fills with peers, the sequential pinging strategy of might cause contention to other pcache operations that the proxy or other users (none that I know of other than proxy) might be requesting. Look into parallelizing the update function so it can run in a timely manner.

[mwyzhao]

@t-lin @hivanco do you guys have any sort of setup for testing I can test the new changes on. It's been way too long since I ran the system and I'm a bit out of the loop.

[t-lin]

@michaelweiyuzhao You can use the VM 10.11.69.7, I've put your public key from client1 in there. The services are in a screen called p2p (attach via screen -x p2p). It's an all-in-one node so bootstrap, allocator, proxy, and registry are all there in different tabs of the screen.

[t-lin]

@michaelweiyuzhao Note that I've just finished #62 and updated the pcache implementation, so hopefully you can pull those changes without conflict.