MySQL Bugs: #1606: allow to "resume" for LONGBLOB

Bug #1606	allow to "resume" for LONGBLOB
Submitted:	20 Oct 2003 5:56	Modified:	11 Nov 2009 2:35
Reporter:	Ralf Hauser	Email Updates:
Status:	Won't fix	Impact on me:	None
Category:	Connector / J	Severity:	S4 (Feature request)
Version:	any	OS:	Any (any)
Assigned to:		CPU Architecture:	Any

Description:
Once one gets into the large sizes of up to 4 GBytes, interrupted communications is not unheard of.
Therefore, it would be great not to be forced to start over, but to be able "resume" a query at any given point!

While this may not be really a part of a JDBC implementation, it would be great to mention it in the corresponding documentation and probably also in http://www.mysql.com/doc/en/BLOB.html

One open source project to mention if it is about post-processing a longblob in a networked environment is probably the Java BitTorrent client
http://sourceforge.net/projects/azureus/ (I didn't test it so far!).

Similarly, BitTorrent concepts may not only be useful due to their "Resume" feature (http://bitconjurer.org/BitTorrent/FAQ.html), but also have some merits when thinking on how to allow for parallel downloads (e.g. if there were a MySQL cluster that does not use the same WAN-gateway).

How to repeat:
.

Suggested fix:
.

Ralf Gebhardt from mysql germany mentions that typically, the application would create chunks of say 1 MB and then keep track of what has successfully been shipped and what not itself. Similarly, it would do the assembly/disassembly of the BLOB from the chunks to the real data.