Add an update_time column to the source table and do the incremental load
by that update_time column.
On Wed, Aug 7, 2013 at 12:04 PM, shengjie min <[EMAIL PROTECTED]> wrote:
> Hi guys,
> TO simplify my question, Let's say, I have a mysql table called 'student',
> looks like this:
> | id | name | sex |
> | 1 | Alice | 0 |
> | 2 | Bob | 1 |
> | 3 | Charles | 1 |
> I want to import this table to HBase periodically which means I will run
> this sqoop job periodically. There are two goals:
> A. every time there is a new record inserted to mysql table, e.g. (4,
> David, 1), I hope my next sqoop import will catch it and put it in HBase.
> B. if there is any updates have been made to mysql rows 1, 2, 3, I want
> to have the updates in HBase too after next round sqoop import.
> I checked two types incremental updates sqoop has: Append mode seems only
> satisfied goal A while Last-modified mode will require my mysql table has a
> timestamp column for each row(which I don't in real life). I know if I
> don't use incremental updates options at all, I can just get way with it by
> running a fresh import every time, but if my mysql table is really huge and
> fresh import might be a performance killer.
> Is there anyway I can just do incremental updates instead of having to
> re-run the whole import to get NEW RECORDS + UPDATES ON OLD ROWS?