Dzero - The fastest globally distributed SQLite database
Fully managed, low latency, high availability, and globally distributed SQLite database
Dzero is a fully managed serverless SQLite database for production scale. Instantly distribute your database on EDGE nodes around the world at no extract cost. Query your database from any environment or framework. Easily manage your database with the table editor.
Access your database from nodes all around the world without having to duplicate replicas and managed them. Every query runs in a secured transaction for data reliability and consistency across all nodes.
Dashboard
Manage your database with a dashboard with all the tools you need
Text to SQL
Generate SQL queries with AI using natural language. Just ask and it will convert it to SQL for you.
Dzero is 100% free during the early access period. Our pricing model will be based on the number of rows read, written, with no additional costs for the number of locations, bandwidth or storage used. We have done up a first cut on what the paid plan might look like by open beta, there might be a free tier too. This will change based on feedback.
Paid plan
$19/mth
Rows read
2 billion included/month + $0.005/million rows
Rows written
4 million included/month + $2/million rows
Text to SQL AI
Not sure yet
Storage
Free (10GB/database)
Replications
Free (320 locations)
Backups
30 days
Logs
30 days
Analytics
30 days
Typically, database providers charge per replication, so if you write 1 row with 10 replicates, you are charged for 10 writes. You will not be charge for replication on Dzero, your usage are only counted for rows read and written regardless of replication, therefore 1 row written to 320 replicates, is only charged as 1 row written.
Database logs can get expensive to store overtime, we are still working on logs but the objective is to keep it free and provide some form of log drain.
FAQ
Technically there is no limit, however we are rate limiting to 300 requests per second per account during the early access period.
For early access there is a limit of 2 databases per account. This limit will be removed and unlimited databases will be allowed in the future.
Unlimited rows, you're capped by size of a database which is 10GB.
It's a hard cap by Cloudflare D1 regardless of tier. The assumption is to stay true to SQLite allowing it to be portable and easily quickly replicated across all edge nodes. Ways around this is to create multiple databases for large tables that don't require joins/foreign key access. We are looking into fully managed vertical partitioning to allow for "infinite" scale.
We take daily backups of your database and store them for 30 days. This feature isn't currently available but will be in the future.
No, regardless of the number of database replicated, locations. You will only be charged for 1 database and not the replications. So if you have a 3gb database, only 3gb will be used out of your 10gb limit and not 30gb multiplied by the number of replications.
Writes are slightly more expensive than reads because they require more resources to replicate across all edge locations. Reads are cached and are faster to serve. We are looking into ways to reduce the cost of writes in the future. This shouldn't affect most users as most applications are read heavy.
Each database is created in an isolated sandbox environment. This means that your database is not shared with any other user and is completely isolated, secure and private. Each database is only accessible by a token generated when your database is created.
Yes, you can use an index to reduce the number of rows read by a query.
Create an account and you will see a form to request for early access with a short description of your use case. We are looking for feedback and use cases to improve the product.
Todo (By priority)
Smart SELECT statement caching
Logging/Query history
Daily backups
Database analytics
Index editor & recommendation engine
Downloading tables as CSV
SQLite functions editor
Better mobile interface
Vertical partitioning for "infinite" scale
Event triggers calling APIs
More prebuilt SQLite functions
Built-in fully managed global vector database (maybe)
If you have a feature request or recommendation, please feel free to reach out!
Who am I?
Hey there! I'm Yoeven and I love the concept of serverless technology that takes away the infrastructure and devops work, allowing us to build things that matter! I have also created JigsawStack, a fully managed commodity AI API service platform which powers aspects of Dzero. You can reach out to me on Twitter or LinkedIn.