tags:

views:

926

answers:

2

There are 3 different ways to get data out of a Blob column: getBytes getBinaryStream getBlob

Also, the Blob object returned by getBlob also has a getBytes and getBinaryStream on it.

Are there any particular reasons (performance, memory, database specific problems) that I should pick one over the other?

The Blob object also has a free() call that has been introduced since JDBC 4.0. Does that make a difference?

+3  A: 

Generally you want to pick the stream-based methods (i.e. getBlob().getBinaryStream() or getBinaryStream()) rather than the byte-array method.

  1. Performance. The driver has a chance to incrementally pull bytes from the database.
  2. Memory. You don't have to load all bytes at once, and in one contiguous block.

Worst-case is the database (or JDBC driver) doesn't truly support streaming binary data, but then there's still no appreciable penalty for using the streaming methods.

Jason Cohen
+2  A: 

If you're going to be pulling a lot of data (i.e. enough data to cause memory problems), then getBinaryStream will give you most flexibility to process and discard the data as you read it in.

On the other hand, this could be quite slow, depending on your JDBC driver, since each read from the stream could entail a lot of network chatter with the database. If you call getBytes, then the driver knows to fetch the whole lot in one go, which is likely to be more efficient.

getBlob() returns a "pointer" to the data, which you can manipulate using the methods on the Blob interface. If you need to modify or otherwise get fancy with the data in-situ, then this might be best for you.

skaffman