d1 Can I have a rough schedule of when this beta or official release will be?
d1 Can I have a rough schedule of when this beta or official release will be?
MWe don't give rough schedules, though we announce things like this at developer week or birthday week
Khi, is it possible to beg for a 3x size increase in D1? I'm very eager to try for a prototype, but my biggest table is 173M in sqlite format so I need at least 300M
D
Hi! The current implementation of D1 is limited by the memory limit of Workers/Durable Objects. The docs say that this limit will be increased in the future, I'd say 
KThe docs also say that if I need more then
If your use case requires a larger database, please reach out!but I got the point, alas...
Dthey are gathering feedback, and looking for use cases so they can make sure to support them in GA
EHi, is this normal for D1 to be so slow? I get that it's alpha, but 500ms+ to do a SELECT on a 230KB database? Doing 2 or 3 more SELECT and you're easily 1.5s+ ... sounds like something's wrong here to me
EAnd with local wrangler, same operation is about 1-2 ms
EWhether thats normal depends on a few factors, like where the client is located compared to where the database was first populated/created in, however it has been said that speed is not something thats been looked at very much in this stage of the alpha, and it will get better in the future
W^ remember that local is local
MHopefully D1 gets faster than querying a DB from another cloud
Nwhen i run
insert into user_agent (value) values ('test2') on conflict(value) do nothing returning idand how should i do this logic efficiently, if this isn't the right way?
Nthis is what happens in case of an empty new table, running the above statement:
1) insert
2) insert
3) insert
1) insert
test12) insert
test3) insert
test13well
insert into user_agent (value) values ('test2') on conflict(value) do update set value = excluded.value returning id
MHave you looked at the docs here? https://developers.cloudflare.com/d1/learning/using-d1-from-pages/
A binding enables your Pages Functions to interact with resources on the Cloudflare developer platform. Use bindings to integrate your Pages Functions …
EI am from Madagascar, and normally there is a Cloudflare Data center here in the capital, but it seems to be not stable at all, or may be incomplete, I have no idea but it's unreliable (sometimes its up, sometimes it's not, and not all services are available there, like there's not WARP). So, the second hop from there is in South Africa (that one is reliable so far). That's about 60ms away.
I assume the DB should be created there. It makes no sense to create it further away.
Now, @Walshy | Pages I agree that local is local, but that should give us a good baseline of how much that operation really costs. With pessimistic estimation, it should be anywhere around 100ms for a single simple SELECT operation, assuming there are much overhead involved, I'd say at most 200ms. Now, 500 - 600ms response time, that's hard to understand why.
While I'm just might skip D1 for this current side project, I hope D1 will live up to the high standards of other Cloudflare products that I really love.
I assume the DB should be created there. It makes no sense to create it further away.
Now, @Walshy | Pages I agree that local is local, but that should give us a good baseline of how much that operation really costs. With pessimistic estimation, it should be anywhere around 100ms for a single simple SELECT operation, assuming there are much overhead involved, I'd say at most 200ms. Now, 500 - 600ms response time, that's hard to understand why.
While I'm just might skip D1 for this current side project, I hope D1 will live up to the high standards of other Cloudflare products that I really love.
HNote that not all Datacenters actually support DOs. As an example, these are the spawn statistics for DOs created from Antananarivo:
MSpeaking of performance, the D1 dash is now optimistically loaded, so you should see significantly fewer loading screens as you navigate around
Mif you're running wrangler to test this, you'll need to add
--persist
Jis anyone else having issues creating new backups?
JConstantly getting 504 Gateway Time-outs, already created a GitHub issues, just trying to figure out if anyone else is having similar issues
JWe've also been getting more and more 9000: Something went wrong errors the past couple of days 
JIt's pretty large, probably larger than most databases. Looking up the exact numbers...
J6.7mb according to the dashboard, but it's probably a lot more. — it has to be a lot more
— it has to be a lot moreJThe database has to be a lot larger, the 6.7mb don't make much sense
JIs there any way to see an exact number? By downloading a backup, I suppose?
JYeah, the backups are going towards the 80mb mark
JProbably, but that must've been a long time ago
JThe database has been growing for a long time now, I haven't imported a bunch of data recently
JHm, I do that pretty often actually
JThe number always seemed a lot smaller than the actual size of the DB, I didn't think it would diverge by 50+mb
JIt would be a valid reason, though
JWhen creating a Backup on the dashboard it says it was created successfully, but it's never showing up in the list.
JSure, I'll DM it to you
ECan anyone explain how D1 handles transactions? I'm trying to wrap my head around whether D1 is protected against two processes attempting to write at the same time
Does D1 detect when a batch/query is submitted whether it contains a write in any of the commands and claims the write lock/transaction? (Sqllite docs say
Cloudflare docs say
Does that mean they follow the sqllite implementation (the first query needs to be a write in order to claim the write lock)? Or is it something custom?
Does D1 detect when a batch/query is submitted whether it contains a write in any of the commands and claims the write lock/transaction? (Sqllite docs say
If the first statement after BEGIN DEFERRED is a write statement, then a write transaction is started.BEGIN...Cloudflare docs say
Our implementation guarantees that each statement in the list will execute and commit, sequentially, non-concurrently. Batched statements are SQL transactions. If a statement in the sequence fails, then an error is returned for that specific statement, and it aborts or rolls back the entire sequence.Does that mean they follow the sqllite implementation (the first query needs to be a write in order to claim the write lock)? Or is it something custom?
EI'm just looking for some clarity as I couldn't see anything official around it in the docs, the comment around
implementationbatches