On 15 August 2018 at 14:12, Wout Mertens <wout.mert...@gmail.com> wrote:
> On Tue, Aug 14, 2018 at 6:28 PM Rowan Worth <row...@dug.com> wrote: > > > FWIW in the building I work in we have 20-30 users hitting around a dozen > > SQLite DBs 8 hours a day 5 days a week, with all DBs served over nfs. > > > Multiple writers? I presume you use WAL mode? > Yes, all clients read and write the DBs. No we don't use WAL mode, as the clients all run on different machines (WAL only works for multiple clients on the same machine). > We plan to migrate to an actual SQL server > > for that reason, but please don't take it as a criticism of SQLite - I > > think it does a marvelous job in a scenario it definitely wasn't designed > > for. > > > > Before you do that, did you see bedrockdb? http://bedrockdb.com/ > (Although it looks like the github repo is being a little bit ignored by > Expensify) > I've given a reasonable amount of thought towards some kind of "server-fied" SQLite backend, but the main problem I'm facing is that our application's DB layer is very coupled to SQLite and its library bindings. The changes/abstractions required for the application to talk to such a backend are the same amount of work to implement as having it talk to an SQL server. Actually replication might allow that work to be bypassed, but there's a lot of unknowns there as to failure modes and how to manage the machines involved in replication as users drop in and out of different DBs in an ad-hoc fashion. Also on the infrastructure side we have a push towards net-booted diskless nodes... -Rowan _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
