an atreus but close enough. You mean .emacs inspiration?

Hello! I am looking for a managed database with the following properties, could anyone point me in a direction? It should be: - decently scalable (not google-scale, but it should handle middle-range load well) - pay as you go - managed - can be sql or nosql, I don’t care too much - doesn’t have to provide incredible persistence guarantees, just something good enough

the pay as you go bit is important, as it should be cheap to operate under small load

I’ve really come to love embedded databases like sqlite

they’re the opposite of managed though

it’s not managed, but it’s very manageable

in some cases it’s easier to manage than “managed” databases

otherwise, it probably just depends on what hosting you’re using

if you’re on aws, then you can use rds

if you’re on google cloud, they have a few options

with DynamoDB you pay per read and write capacity, so that could be an option

@schmee DynamoDB was the only option I could think of. RDS as suggested by @smith.adriane isn’t very granular in its billing; you pay per instance

I don’t see how I would use Sqlite if I am running more than one instance of the application

sqlite is not a good option if you have more than one server

I would love Google Cloud (which is what is taking me away of DynamoDB a bit) but none of their offers seem to fit the bill except cloud datastore

but it seems a bit outdated

I think aws aurora on rds is more of a pay as you go model

>Storage consumed by your Amazon Aurora database is billed in per GB-month increments and IOs consumed are billed in per million request increments. You pay only for the storage and IOs your Amazon Aurora database consumes and do not need to provision in advance.

from https://aws.amazon.com/rds/aurora/pricing/

@hmaurer have you looked at BigQuery or Spanner?

with BigQuery you pay per query, dunno what the model for Spanner looks like

@schmee those seem to be priced per instance at start at a very high cost. I am trying to build a cheap architecture for a non-profit educational app. I was thinking of using Cloud Functions to scale on that side, but the database is problematic (I can’t afford upward of 100$ costs per month for no traffic)

you sure? I can’t find any per instance costs for BigQuery: https://cloud.google.com/bigquery/pricing

for cheap options, I’ve used the $5/mo instance on digital ocean using server+sqlite

@schmee I am not familiar with BigQuery; it seems to be aimed at big data but it might be suitable for my use-case too

i’ve also used heroku in conjuction with the free tier offerings for their plugins

specifically, clojure + redis

which I’ve had a website running for years

it’s limited to 25MB of data

but it’s been running for free

@smith.adriane that’s what I did last year (with digitalocean), but it doesn’t offer a nice path to scaling to multiple instances without managing it myself. I could set up a kubernetes cluster on digitalocean or google cloud and manage my own database though… but it has an upfront cost

does heroku have persistence for redis?

ah yes it does

mmh

just looked at my account. it looks like I was using https://elements.heroku.com/addons/rediscloud

yeah actually no way I can use Heroku’s redis offering as a database; it’s prohibitively expensive storage-wise

120$/month for 500MB

yea, i’m not sure what kind of load/pricing you’re targetting

could use their postgres though, but that would have the same issue as RDS / google cloud sql

ideally I would like a pay-per-write/read, like DynamoDB

or pay per GB stored

one slightly off-beat option is to use S3 + Athena

S3 costs jack shit for storage, and Athena costs per data scanned per query

@schmee yes I was actually just looking at this right now (but with Google Cloud Storage instead of S3)

it seems like a viable solution to store JSON documents

a bit unorthodox, but heck…

if keeping the costs down is a priority then it is a very viable option

@schmee basically, to give you an idea, I am writing an education application for a course at my universitity, which we would like to open up to other universities

we need to be able to onboard new universities for free (to encourage adoption), and so the costs for us should be low, and smoothly increase (thus pay-as-you-go)

the amount of data could be decently large

the course we will be using it on at our university has 400 students, but we would be storing a lot of exercise data and student solutions

so it could grow to a couple of GB over the course of the semester

and that’s only a single course

json blobs on CLoud Storage / S3 seem like a decent solution actually

there might not even be a need for Athena if I structure the data properly

I don’t really need ad-hoc querying

being able to store and retrieve JSON documents quickly should be enough

I am not sure what the latency of S3 is if I query it internally (within AWS’s network)

but I would hope it will take < 20ms to retrieve a small json doc

I just checked and google cloud storage supports optimistic locking, which is good too

neat

latency should be good enough

it is the easiest way to get up and running too

have you heard of people using S3 as a database before? just to be sure it’s not something completely stupid to do 😄

yeah

people use S3 for anything and everything

google cloud storage charges more per write operation

but it still cheap it seems

0.05$ per 10,000 writes

just stumbled over this too: https://aws.amazon.com/simpledb/

how about Firebase?

Thank you @schmee 🙂

@sundarj yeah I am considering that one too… It seems like a decent option; I need to check how much data I will have (in GB) as firebase can get quite expensive on that side