This is an amazing feature for when you have not-so-important data that's quickly accumulated and that you need to do some data analysis on (which SQL is really useful for).
I commented over there on the blog that I could easily see a generic solution that moves data from unlogged into logged tables at a set interval, helping to keep request time low (by inserting into the unlogged table), helping to keep i/o lower (by bulk inserting into the logged table which causes only one WAL write and thus fsync) and still providing a reasonable amount of data integrity.
For my personal use-case, this is one of the best features I've ever seen appear in PostgreSQL and I can imagine it could be of some use for other people here, hence the submission.
I commented over there on the blog that I could easily see a generic solution that moves data from unlogged into logged tables at a set interval, helping to keep request time low (by inserting into the unlogged table), helping to keep i/o lower (by bulk inserting into the logged table which causes only one WAL write and thus fsync) and still providing a reasonable amount of data integrity.
For my personal use-case, this is one of the best features I've ever seen appear in PostgreSQL and I can imagine it could be of some use for other people here, hence the submission.