Currently when syncing a bitcoind node from scratch the time of going from zero (0) to about 50% is a very fast process but somewhere between 50%-75% it slows down considerably and at 75%-85% it grinds to an almost halt.
Assuming that the blocks that are being verified during that time are what is causing the slowdown I'm wondering if there is a way to basically clone a node by still running and validating a full node but at least trusting the blocks up until some number that is far enough in the past to handle 99% of edge cases.
I know that cloning the disk is possible but transferring that to a new VPS for instance is not so looking for ways to speed up the process for everyone without sacrificing security.
Known Optimizations
Setting the dbcache
to a higher value (currently using 2000
on a 4gb RAM machine which is 25% of the available RAM)