When will cloudflare announce they have abandoned read replicas? It’s been years now, and it doesn’

When will cloudflare announce they have abandoned read replicas?

It’s been years now, and it doesn’t help that the blog post back in April talks about replicas as if they are already deployed, it’s the first result shown when you search for D1 replicas. Not a great look to be dangling this carrot for so long and provide so few/zero updates for nearly a year

Ashkan•1/25/25, 8:44 PM

Dear CF team I think Console Tab needs some improvements. in addition to notes in the picture. there some suggestions; 1- Use different background color for query input area. 2- it's better to not execute queries on ENTER Key Press Event and execute it on Mouse Hover and Mouse Click on Execute Button. it prevents on unwanted ENTER key press maybe for going to next line. (Personal Opinion)

Mmatt When will cloudflare announce they have abandoned read replicas? It’s been year...

1984 Ford Laser•1/25/25, 10:23 PM

There have been posts more recently than March mentioning replicas, and employees are in here mentioning it from time to time. It is not ready, and won't be released until it is. Feel free to move onto another suitable product, if this one doesn't suit you

11984 Ford Laser There have been posts more recently than March mentioning replicas, and employee...

mattOP•1/25/25, 10:25 PM

where can i find any updates provided since that blog post nearly a year ago? I've searched everywhere, i've searched for "replicas" in this discord and scrolled back through 6 months of messages

mattOP•1/25/25, 10:25 PM

has anyone heard of peter molyneux?

Mmatt has anyone heard of peter molyneux?

1984 Ford Laser•1/25/25, 10:26 PM

Is this relevant? There is an #off-topic

mattOP•1/25/25, 10:28 PM

its relevant with regards to the relationship between promising & delivering

1984 Ford Laser•1/25/25, 10:28 PM

Okay well it isn't ready, that's the update.

mattOP•1/25/25, 10:29 PM

gosh i feel for the developers, it must be a lot of pressure putting out announcements well before things are ready. i imagine it feels a bit like the crunch phase of game development

mattOP•1/25/25, 10:30 PM

reminds me of when i used to work in big orgs and the sales team would promise new customers that we had features which weren't built yet... so the PMs would come down on us hard to build something we didn't get the time to plan properly

Althaf•1/26/25, 12:50 PM

do anyone have any idea why i can't acess the d1 using demo as i have local d1 in .wrangler file and still this error occurs like the env.db is undefined like does the d1 doesn't work on server actions ? nextjs cloudflare pages

Althaf•1/26/25, 12:51 PM

In nextjs

AAlthaf do anyone have any idea why i can't acess the d1 using demo as i have local d1 i...

森

森@VerseLink (RW)•1/26/25, 2:33 PM

you should try to see if not using drizzle works first

森

森@VerseLink (RW)•1/26/25, 2:33 PM

if so, then potentially it is with your drizzle config

森森@VerseLink (RW)you should try to see if not using drizzle works first

Althaf•1/26/25, 2:39 PM

Oh thats a good point

森森@VerseLink (RW)if so, then potentially it is with your drizzle config

Althaf•1/26/25, 2:44 PM

my config is like this. i think the issue is like when i try to console.log env.DB or env.DATABASE both are undefined i have no idea why is that and i explicity add the file path still error occurs like } db.query.users
⨯ [TypeError: this.client.prepare is not a function] {
digest: '2881251540'
} is there any fix for it or like does d1 doesn't works on server actions ?

AAlthaf my config is like this. i think the issue is like when i try to console.log env....

森

森@VerseLink (RW)•1/26/25, 2:47 PM

what does your wrangler.toml look like

Mmatt When will cloudflare announce they have abandoned read replicas? It’s been year...

Max (@rozenmd)•1/26/25, 4:16 PM

it is actively being worked on, lol

森森@VerseLink (RW)what does your wrangler.toml look like

Althaf•1/27/25, 4:55 AM

i copied this when generating local d1

LLeesy Hi lambrospetrou, is there any more information on the read replication coming? ...

lambrospetrou•1/27/25, 10:08 AM

Folks, someone asks for replication every day. When we have more information to share publicly, we will share it here and make it noticeable for everyone. It's been actively worked on. We cannot give more details on release date now.

Llambrospetrou Folks, someone asks for replication every day. When we have more information to ...

Hello, I’m Allie!•1/27/25, 10:43 AM

Well, there is always

Hello, I’m Allie!•1/27/25, 10:43 AM

:soontm:

Rudi B.•1/27/25, 1:00 PM

Hello, any1 know If it is planned for the Feature to öfter Mongo DB Databases?

RRudi B.Hello, any1 know If it is planned for the Feature to öfter Mongo DB Databases?

Isaac McFadyen•1/27/25, 1:04 PM

It's not planned. MongoDB is a very very different architecture to D1 and there are no plans to make D1 Mongo compatible that I am aware of.

Stefan Stührmann•1/28/25, 3:50 PM

Hey, are there any plans to provide the possibility for data analytics. As far as I understand I can access D1 either via worker or via the proprietary http api. Sure someone can build an own etl solution around this but this requires own development efforts (and you have to take care about change data capture etc. yourself). Also you also have to break out of the cloudflare eco system as there is no analytics optimized data store. A good start would be to have the possiblity to etl d1 / kv etc. to R2 -> iceberg or managed Clickhouse (which is already there for workers analytics).

SStefan Stührmann Hey, are there any plans to provide the possibility for data analytics. As far a...

Max (@rozenmd)•1/28/25, 4:20 PM

change data capture is something we'd like to offer with D1, yeah

Robin Genz•1/28/25, 4:27 PM

Hey

Quick question: Is there any way to increase the 10 GB limit of Cloudflare D1 for my account? At the moment it is not a problem but we are growing very fast and will reach the limit in the next few months.

MMax (@rozenmd)change data capture is something we'd like to offer with D1, yeah

Stefan Stührmann•1/28/25, 4:31 PM

how would you implement it cdc -> r2 with append operation like S3 got recently or you plan to introduce a stream like Kafka?

Original message was deleted

iLearn•1/28/25, 9:59 PM

Is the hard-limit of 10GB also applicable in case of DO with "SQLite storage" (which is D1) ?

Original message was deleted

iLearn•1/28/25, 9:59 PM

Thanks for prompt and crisp answer.

Original message was deleted

iLearn•1/28/25, 10:00 PM

Agree.

Original message was deleted

iLearn•1/28/25, 10:01 PM

True. But, can be raised to 10GB on request and ofcourse, in GA it will automatically be raised to 10GB.

rufi•1/29/25, 3:20 PM

Say I read from D1 on every request for some config or the like, would this add something like 200ms latency on each one?

Rrufi Say I read from D1 on every request for some config or the like, would this add ...

lambrospetrou•1/29/25, 3:59 PM

it depends where your database is, and where your workers will be. Usual roundtrip latencies apply. If you query from Europe to a database in US 100-150ms is expected yes. Within the same region, like EEUR/WEUR you can get 20-60ms.

gaminski•1/29/25, 4:55 PM

I need help understanding something. On the D1 marketing page it states we can have up to 50,000 databases on the paid plan. I cannot find a way to bind multiple databases to one project. How is that done? I’ve searched this thread and I realise that dynamic binding isn’t yet a thing. Some people mention binding multiple databases to one worker but I’m not sure how that’s done

Devid•1/29/25, 4:57 PM

Hello, Is there any way to migrate aws rds mariadb database to cloudflare database?

Ggaminski I need help understanding something. On the D1 marketing page it states we can h...

lambrospetrou•1/29/25, 5:23 PM

You can have multiple D1 bindings in your

wrangler.toml

wrangler.toml

. Repeat the

[[d1_databases]]

[[d1_databases]]

section for each of your databases. That's defining an array, and each time you use it, it's an item in that array. The newer

wrangler.json

wrangler.json

format makes that a bit easier to really understand though if you are not familiar with TOML syntax.

Example:

[[d1_databases]]
binding = "DB_01"
database_name = "db-01"
database_id = "<unique-ID-for-your-database>"

[[d1_databases]]
binding = "DB_02"
database_name = "db-02"
database_id = "<unique-ID-for-your-database>"

[[d1_databases]]
binding = "DB_01"
database_name = "db-01"
database_id = "<unique-ID-for-your-database>"

[[d1_databases]]
binding = "DB_02"
database_name = "db-02"
database_id = "<unique-ID-for-your-database>"

DDevid Hello, Is there any way to migrate aws rds mariadb database to cloudflare databa...

lambrospetrou•1/29/25, 5:24 PM

Export/SQL dump your existing database and import it in D1: https://developers.cloudflare.com/d1/best-practices/import-export-data/

Llambrospetrou You can have multiple D1 bindings in your `wrangler.toml`. Repeat the `[[d1_data...

gaminski•1/29/25, 5:25 PM

Thank you - that sounds simple! Is there a way to limit access to a db to specific subdomains in the config file or elsewhere on the worker, or does it need to be done at the application level?

Ggaminski Thank you - that sounds simple! Is there a way to limit access to a db to specif...

lambrospetrou•1/29/25, 5:25 PM

The worker that has this binding has full access to the database. If you have code that you don't want to have access in some of your bindings then it should be in a different worker.

Llambrospetrou The worker that has this binding has full access to the database. If you have co...

gaminski•1/29/25, 5:28 PM

Ok thanks. I think im struggling to grasp some fundamental concepts. Can I have multiple workers for one project? Maybe what I’m looking for is one worker per database? I’m trying to allocate one database per user (I’m not expecting thousands of users so I’m happy with a manual solution for now)

Ggaminski Ok thanks. I think im struggling to grasp some fundamental concepts. Can I have ...

lambrospetrou•1/29/25, 5:30 PM

Number of workers and how you allocate each database to users is very different. Your Worker code is the one that has access to the databases, so your code controls which database to access. So you can have for example 10 D1 databases attached to your worker, and then depending on the request context (e.g. user ID, account ID, or whatever your user identifier is) you decide which of those databases to use.

You definitely don't need 1000 worker projects!

gaminski•1/29/25, 5:38 PM

Thank you. Let’s see if I understand correctly: I have one worker with multiple databases. One TOML file with the bindings defined as you explained earlier. Somewhere in my application I keep a record of which database each user should be able to access and route the queries accordingly

Ashkan•1/29/25, 5:40 PM

AAshkan Click to see attachment

Ashkan•1/29/25, 5:40 PM

rufi•1/29/25, 6:49 PM

ok but I assume D1 is geo-distributed. So If my worker uses smart placement, it would reduce latency between the worker and the D1 databases?
Also, if I need a better latency (something like 20 ms tops), should I rather use KV, or D1 backend on durable objects, ...?

Rrufi ok but I assume D1 is geo-distributed. So If my worker uses smart placement, it ...

lambrospetrou•1/29/25, 7:05 PM

D1 is not yet replicated. It's in a single location.
With Worker smart placement, your worker will be running close to the database assuming you have enough traffic (100+ requests per day) for smart placement to do its magic.

If you need 20ms tops from anywhere in the world there isn't much you can do with ANY database. You need to have replicas all over the place. Either you do them yourself, or wait for D1 replication. Workers KV has very different semantics from D1 or DOs.

Llambrospetrou D1 is not yet replicated. It's in a single location. With Worker smart placement...

rufi•1/29/25, 7:14 PM

ah ok interesting thanks a lot. So when I create the D1 database I can also select a location. Most of our users are in NA anyway.
The use case I have is basically I want to "in memory sync" the configuration to be used when some endpoint is called. That configuration has to be updated say every hour but I want to avoid having to redeploy the workers to embed the config in place.

Rrufi ah ok interesting thanks a lot. So when I create the D1 database I can also sele...

lambrospetrou•1/29/25, 7:29 PM

If you rarely update that config and you have many requests needing it, then you can have D1 as the source of truth where you write your config updates, and then also put them in Workers KV. Your worker can read from Workers KV when a stale value (up to 60s) is OK for that case, and go to D1 if you need the latest/freshest configuration.

Using the local CDN Cache API from your worker sometimes give you faster lookups than Workers KV as well, but it also depends on your scenario. It's not clear from what you said above.

This design can give you the speed when you can use cached values and have D1 as a fallback for your endpoints that need the latest.

Ggaminski Ok thanks. I think im struggling to grasp some fundamental concepts. Can I have ...

Larry•1/30/25, 12:24 AM

"one worker per database" sounds exactly like Cloudflare Durable Objects

[[d1_databases]] binding = "DB_01" database_name = "db-01" database_id = "<unique-ID-for-your-database>" [[d1_databases]] binding = "DB_02" database_name = "db-02" database_id = "<unique-ID-for-your-database>"

When will cloudflare announce they have abandoned read replicas? It’s been years now, and it doesn’

Similar Threads

Similar Threads

Similar Threads