jasonmp85
commented
8 years ago Ok, so I edited this for accuracy (I used concurrent where we could, instead of parallel) and added a section about why concurrent queries are a good idea (if we have to tell our users to do this, they likely don't understand the underlying cause). I also rewrote the last bit about psql to be about applications generally (i.e. if you have an application doing writes, figure out how many copies (threads, connections, processes, actors, whatever) of it you need).
Additionally, I removed the use of find, which I consider one of the worst-designed commands in Unix. ls does just fine there… and I removed the echo and sh in favor of -t and -I%. Finally, I mentioned that only GNU split is capable of -n (BSD split doesn't have that option).
But of course the build is failing because PostgreSQL 9.5 has a new beta which changes certain outputs. I'll fix that in develop real quick and force a rebuild here.
This PR updates the README about increaing the INSERT throughput on pg_shard cluster. In the last pg_shard sync meeting, we decided that mentioning the concurrent INSERTs may help users to get better performance.
@jasonmp85 Feel free to change the wording&structure in the PR, my aim is to kick off this change.