On Thu, 11 Apr 2019 18:45:01 -0600 Warren Young <war...@etr-usa.com> wrote:
> Sure, but what *is* on the disk after a crash is always consistent > with ZFS, so any decent database engine can recover. It's been some years, but I saw a presentation about running Postgres on ZFS. Every "victory" was a way to compensate for a feature of ZFS. None of the problems surmounted are presented by normal filesystems. The simple fact is that ZFS does nothing for the DBMS. As you acknowledge, it does complicate matters. And the features ZFS supplies slow down the DBMS by stealing I/O cycles that the DBMS could otherwise use productively. > > It's really not the ideal substrate for a system that takes its > > fsyncs seriously. > > You know, I?ve just realized that it?s been a really long time since > I?ve heard anyone seriously talk about running databases on raw > storage. I don't know what you mean by "raw" storage. A storage substrate that really syncs when it's told to and did do when it says it did doesn't have to be a locally attached SATA drive. It could be a SAN. It just has to be a yankee at the end of a dirt road: simple and honest. ;-) > I?ve even heard of people successfully using ZFS snapshots to make > live, continuous DB replications from one site to another for fast > failover. No doubt. But absence of proof is not proof of absence: just because it seems to work doesn't mean it can be depended on. If the DBMS developer can't vouch for it, it's a sure bet it's depending on at least one fortuitous coincidence. > It calls into question how important, relatively speaking, > lack of mediation is in system storage design. We mustn't place much faith in common (a/k/a "best") practices, viz: > Of course raw storage isn?t the main alternative to ZFS. It?s LVM+md > +XFS and similar lash-ups, which are even worse in this regard. Exactly. The simple understanding that a DBMS has many of the attributes of an operating system, and therefore needs direct access to the hardware, is lost on many (I would say most) system administrators and not a few database administrators. Now if you'll excuse me I have some kids to chase off my lawn. --jkl _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
