Spring集成：JDBC入站通道适配器

Question

I was trying to use JDBC inbound channel adapter and I encountered some confusions.

First, which one is more efficient to use: specifying max-rows-per-poll or using limit on query statement? Second, to avoid the duplicate readings, is there anyway to run poller instead of fixed-rate after last message of each poll is processed? I could use update but there is dynamic update query down the stream.

My integration flow is somewhat like jdbc:inbound-channel-adapter -> http:outbound-channel-adapter -> jdbc:outbound-channel-adapter (update each row as per response from http:outbound-channel-adapter ). It should handle huge number of rows.

Answer 1

1. LIMIT is more efficient.
2. As long as you use DirectChannel s (the default), and no task executor on the poller, the next poll won't run until the current one completes, regardless of the trigger.