DB connection times

Dr. Hawkins dochawk at gmail.com
Tue Mar 3 00:20:17 EST 2015


On Mon, Mar 2, 2015 at 8:21 PM, Richard Gaskin <ambassador at fourthworld.com>
wrote:

> Local DB performance like that makes a good case for working with text
> files. :)
>
> How many records are in there?  Complex indices?  What could account for
> so much time to connect locally?
>

It *is* entirely text.


"remote" postgres database has a single table, full of insertion commands
into three tables.  One has ~ 500 rows with about 7 or 8 columns, another
has a few dozen with about 6 columns, and another typically a couple of
dozen with something like 50 columns.

I've seriously reduced the number of transactions, but there are still
several separate db queries to get a file open and sanity checked, and to
bring in a few but small tables.

It's the latency that's expensive for me; the queries are cheap.  For a
fairly complicated debtor, the gzipped list of commands is 24kb in my
backup file; 16kb is more typical.

The remote query, however, going from my N-wireless to airport expres to
cablebox, then across town on cox to office, into cablebox, to airport
express wirelessly to my iMac seems to typically take 200-250 ms.

Now, when I was doing that on exiting every field, over clear wireless
internet to runrev servers, it was more like 1-2 *seconds* for a
transaction . . .

Thus my interest in going asynchronous with sockets . . .


-- 
Dr. Richard E. Hawkins, Esq.
(702) 508-8462



More information about the use-livecode mailing list