I've read some places that its faster to synchronize from the network than to download and then import a raw blockchain file. Why is it faster to synchronize from the network?
Add a comment
|
1 Answer
Mainly it's a question of available RAM and I/O bandwidth. When you're importing a raw file, reading the raw file takes up disk bandwidth as well as RAM in the filesystem cache. This competes with the disk bandwidth (and RAM) needed to write out the database file. When you're syncing over the network, this extra disk and RAM overhead isn't present.
Note that you can get some of the speed back by using a separate device for the raw file (separate from the device holding the database file). And if you have enough RAM to hold both the raw file and the entire database file, then the import can be done with no speed penalty at all; it will in fact be faster than syncing over the network.
