I use pg in my raspberrypy 3b, for a weather station and works fine, but to be honest, I have a sencond raspi with a pgbouncer and a pendrive of 256GBs as data storage
In other, I have a zabbix server with pg monitoring aboutn 100 devices, but with an external hdd.
but, at the end of the day, I think is not the best idea.
On 2020-10-05 06:20, Thorsten Schöning wrote:
Hi all,
TL;DR:
Does anyone actually use Postgres with ARM based low performance
hardware and only 256 MiB of RAM? What are your experiences in other
stripped down environments? Is there some lower RAM limit with which
using Postgres doesn't make any sense anymore? Is Postgres able to
compete with SQLite regaridng lower performance boundaries or is there
some point at which SQLite will be the better choice, because Postgres
needs some more resources to work properly?
Thanks!
Background:
I have some server side backend exposing web services and storing data
in Postgres. The stored data are telegrams from smart metering
devices, lots of them, mostly small. The provided web services take
care of generating reports etc. based on those telegrams.
It's now considered to put large parts of that backend onto some ARM
based IoT device, simply because that device would do a lot of the
same things. The available backend is Java-based, so it would simply
be possible as well and I already ran things on some Raspi-like
hardware and things worked pretty well. OTOH, the targetted hardware
is less capable than some Raspi, we talk about only 256 MiB of RAM
currently. The exact details of the hardware are not clear yet.
In the easiest case, the newly created device would only need to store
some hundreds of telegrams per day in some known time window and
forward those telegrams afterwards. In most cases it would be those of
the same day only, but might be those of the last X few days as well
in case of problems. In the long term, though, that device should be
able to store telegrams for years and might generate reports on it's
own as well.
I've already found some interesting infos like the following, but
would be interested in other expereinces as well.
https://www.postgresql.eu/events/pgconfeu2019/sessions/session/2781/slides/251/PostgreSQL%20for%20IoT.pdf
Mit freundlichen Grüßen,
Thorsten Schöning
