3 ms·
Is there a push mechanism in DNS? My uninformed model is that it's all pull based, and for changes to propagate you basically have to wait for cached entries t
by bagels 4y ago
Is there a push mechanism in DNS?
My uninformed model is that it's all pull based, and for changes to propagate you basically have to wait for cached entries to expire. It seems to me that if there were a cache invalidation mechanism you could have fast updates, but long cache TTLs.
(I know that if this doesn't exist, adding it would be basically impossible, since some servers won't update to new protocols)
- School-Cotton 4y agoHow would that even work, since you can’t in general know who has cached the entries you control? Broadcast updates to the whole internet?
- wnoise 4y agoThere is the NOTIFY extension, but it's for master servers to notify secondary servers that they should do a pull. Not really usable for unknown general clients, of course.
- wahern 4y agoThere's DNS NOTIFY (RFC 1996), which is commonly used to trigger a pull. But to minimize malfeasance it's usually ignored unless authenticated (e.g. via DNS TSIG) or appears to originate from a known address. The typical scenario is a master nameserver notifying a list of [known] slaves about a zone update. There's also DNS Push (RFC 8765), but it's a very recent mechanism. I'm not familiar, but I doubt it's widely supported.