On 11/30/08, Brian Aker <[EMAIL PROTECTED]> wrote: Well, IMO, there are indeed good tools external to the database to do >> all this stuff. Exchanging data through a storage engine in this >> manner would merely be a convenience feature. >> > > Unless it was for performance? Loading data is something that is a low > hanging fruit for us to do. Though in the end it is very dependent on the > engine.
1) the only reason to keep LOAD DATA INFILE is performance. Otherwise, you can use batch insert commands. 2) I think working on how loading data in is done is very important to do. 3) I remain unconvinced this has to be done in the core microkernel. For the folks using LOAD DATA INFILE, how many actually need the performance? How many do it regularly, and why? For example, if a slave can have more than one master, does the need for LOAD DATA INFILE go away? What about some kind of Oracle->Drizzle plugin, or anyDB->Drizzle? Does that negate the need to run the LOAD DATA INFILE frequently? So far from the examples there's exactly 1 explicit case (from Ask) of LOAD DATA INFILE being used regularly (ie, not just for setting up a table the first time) -- and we don't know the original data source, nor what it's used for. If Drizzle is going to be used for, say, data warehousing, people can write plugins to do this kind of ETL quickly. In "random aside tangents", though, I'd love for there to be a plugin where something like LOAD DATA FROM [server] can set up a slave from a master with very little locking. -Sheeri
_______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
