Embed Notice

@futurebird
Your idea was implemented for the leap second: The Google leap smear.

Instead of handling leap seconds in all their code, Google had their time servers fall half a second behind slowly over 12 hours. Then after the leap second, they were ahead by a half sec, which they caught up to over the next 12 hours. That way they kept all their servers in sync without the dreaded step change, and apart for 24 hours (+a second), also in sync with everyone else.

http://wordpress.mrreid.org/2013/04/23/leap-smear/