Is druid suitable for processing mutable data?

Hi all,

I’m trying to anylysis mutable data in realtime with druid. My primary datastore is mysql. Say, it has a table named order which store some order infomation. The problem is the order status may be changed in few hours.

For example,currently,the “status=1” order num is 8000,two hours later, the “status=1” order num may be 10000.

I want to use druid anylysis the order num with diffrent status in realtime. Is it suitable in such scenario?

What’s the best solution for this?

Thanks in advance!

Hi,

If you are constantly appending new events in your data store, then Druid is a good solution. If you constantly updating the same set of events in the data store, Druid is not. Druid is meant for OLAP workflows, and not OLTP workflows. Is order number a count in your table or a field that is edited?

Yeah, in my “order” table the “order_status” field may be updated upon the customer pay for the order in several minutes or hours.
I just want to anylysis how much orders are paid and how much orders are not paid in realtime .

在 2015年7月22日星期三 UTC+8下午1:16:47,Fangjin Yang写道:

Hi,

With a data store like Druid, you should just insert every order event you have into the table. Druid will automatically figure out how to compact rows and update measures such as order_status. This doc may provide some more info:

http://druid.io/docs/latest/design/