NGenerally always feel cloudflare is developer friendly with it's pricing but was up for a huge surprise looking at d1 pricing
It says zero egress fee but charges
0.75$ per gb for storage
where as other providers like firebase and supabase although egress is not free, feel are charging way less,
For example this is the supabase pricing
Storage
then $0.125 per GB
Bandwidth
then $0.09 per GB
Am I missing something or does that mean d1 is more appropriate for warm data
It says zero egress fee but charges
0.75$ per gb for storage
where as other providers like firebase and supabase although egress is not free, feel are charging way less,
For example this is the supabase pricing
Storage
then $0.125 per GB
Bandwidth
then $0.09 per GB
Am I missing something or does that mean d1 is more appropriate for warm data
CI'm not sure you can compare either of those really directly, supabase has their "monthly active users" abstraction and firebase is nosql.
D1's pricing is more against other SQlite providers like Turso: https://turso.tech/pricing, which you'll find their pricing suspiciously familiar.
For example vs Firebase
Firebase (w/ documents logic and not rows):
$0.18 GiB/month
0.6 US$/per mil reads (docs)
1.8 US$/per mil writes (docs)
D1:
$0.75 GB-mo
0.001 US$/per mil reads (rows)
1.00 US$/per million writes (rows)
Barring in mind the big difference between rows read in a sql db and documents read in a nosql db, Firebase is $600 per billion documents read, and D1 is $1 per billion rows read. Indexes do add at least one row written on every row update, but of course could potentially way offset the read.
Taking away "D1 charges more for storage so I should use less/throw big blobs into R2/object storage" isn't a bad take away though. Worth mentioning D1 is 10 GB max per Database, so certainly way more limited
D1's pricing is more against other SQlite providers like Turso: https://turso.tech/pricing, which you'll find their pricing suspiciously familiar.
For example vs Firebase
Firebase (w/ documents logic and not rows):
$0.18 GiB/month
0.6 US$/per mil reads (docs)
1.8 US$/per mil writes (docs)
D1:
$0.75 GB-mo
0.001 US$/per mil reads (rows)
1.00 US$/per million writes (rows)
Barring in mind the big difference between rows read in a sql db and documents read in a nosql db, Firebase is $600 per billion documents read, and D1 is $1 per billion rows read. Indexes do add at least one row written on every row update, but of course could potentially way offset the read.
Taking away "D1 charges more for storage so I should use less/throw big blobs into R2/object storage" isn't a bad take away though. Worth mentioning D1 is 10 GB max per Database, so certainly way more limited
NFirebase let's us do it from client where as here it's hit a cloudflare worker get back the data..so kind of the reads of d1 are cloudflare worker invocation plus db reads..
So for 1M reads+ 1M worker invocations(well a select all is a single worker invocation but multiple reads) = 0.3$ additional for worker
It would be nice if you can allow something similar to firebase let's users use d1 directly from client
(without needing any worker to write, i.e you taking care of workers part automagically) plus with offline capabilities.. the pricing mark up surely makes sense..
to handle Authorization carefully here..supabase has RLS and zenstack and remult offers a approach of integrating them into schema itself but it's away from sqlite
So for 1M reads+ 1M worker invocations(well a select all is a single worker invocation but multiple reads) = 0.3$ additional for worker
It would be nice if you can allow something similar to firebase let's users use d1 directly from client
(without needing any worker to write, i.e you taking care of workers part automagically) plus with offline capabilities.. the pricing mark up surely makes sense..
to handle Authorization carefully here..supabase has RLS and zenstack and remult offers a approach of integrating them into schema itself but it's away from sqlite
