On Fri, Sep 10, 2021 at 07:24:19AM -0400, Timothe Litt wrote: > Clearly map format solved a big problem for some users. Asking whether > it's OK to drop it with no statement of what those users would give up > today is not reasonable. > > After all the "other improvements in performance" that you cited, what > is the performance difference between map and the other formats?
In recent benchmarks, map loaded about 1.5x faster than raw. When we first implemented it in 2014, it was coming out closer to 4x faster. I suspect what's happened is that improvements to memory management, and perhaps also changes in hardware used for benchmarking (SSD vs spinning disk, for example?), have eroded the advantage that map had over raw - in other words, it's not that map's slower now, but that raw has gotten faster. And I'm not sure we would have chosen to incur the complexity cost of the map format if it had only been a 1.5x speedup at the time. Map has always been fragile - you can't copy files from one machine to another, and it doesn't even reliably keep working on the same machine when you upgrade BIND. It's always been recommended only for secondary zones, so that if you upgrade BIND and your old map files stop working, it can automatically retransfer the zones. Using map for primary zones is possible but requires a lot of caution when upgrading or migrating; I don't think very many people do that. Recently a critical bug was discovered in which map files that were generated by a previous version of BIND caused a crash in newer versions. It took over a month for anybody to report the bug to us, which suggests that the number of people willing to put up with such a finicky format must be pretty small. (Or that the people who use it aren't keeping up with regular software updates, I guess.) So, this thread is to ask if anyone is currently relying on map format for reasonable startup speed, and if so, could you compare it against raw and see if you still really need it? If the feature is not being used to any significant degree then it would be good to simplify. It would also make it easier, maybe someday, to replace our database structure with something more performant. (Not that we have any specific plans for that, but it's something we talk about occasionally, and it would be nice not to have to worry about map files when it came to maintaining feature parity.) -- Evan Hunt -- e...@isc.org Internet Systems Consortium, Inc. _______________________________________________ Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe from this list ISC funds the development of this software with paid support subscriptions. Contact us at https://www.isc.org/contact/ for more information. bind-users mailing list bind-users@lists.isc.org https://lists.isc.org/mailman/listinfo/bind-users