• Decision made by a developer before him
  • Started about 12 months ago
  • Schema uses older syntax
  • GraphQL API in front of EdgeDB
    • Using their own GQL API, not EdgeDB GQL support
  • Not doing anything with EdgeDB they couldn’t do with a traditional database
    • Lack of exploration into what it could do for them, USP
  • Using EdgeDB Cloud
    • Elliot appreciates the UI
    • Would like to export results of large query to file
      • CSV or JSON
    • Copy sometimes fails for large results
  • Dane loves the CLI
  • App hosted in Vercel
    • Looking at migrating to CloudFlare
      • Vercel is too expensive
      • No need for much of what Vercel offers
      • Loves CloudFlare Workers and excited about Pages
      • Yury will ask for discount on team membership
  • High Availability: Support is upcoming. We will automatically route read-only queries to the closest database you have deployed.
    • We own the whole stack, which is relatively unique for databases. We own protocol, query language, DSL, client libraries, etc.
    • Sharding will be defined in your schema
    • Dane: they are experiencing some latency and would be interested in HA
    • Sharing will cost extra on top of existing tiers
      • Existing tiers will not be raised as part of this, but you will pay for regions you add
  • Yury and Elvis want to stay close to small companies because they drive innovation
  • Elliot: Will we have more integrations with Vercel?
    • Yury: They are our primary investor, and we are aligned with them. We are thinking of building some infrastructure for React Server Components that Vercel is betting on. Big announcement in a couple of months.
  • We will automatically provision a branch on our cloud for preview deployments on Vercel
    • Cloudflare will eventually have this integration, but Vercel is asking for it now.
  • Dane: They use RDS PG to back their staging instance. Is there a better way?
    • Yury: You could use a free tier instance.
  • Sane: Social media adjacent
    • Pinterest mixed with Tumblr
    • Looking into discoverability
      • Company will build recommendation engine for them
      • Will get back a list of IDs from recommendation engine, so that query gets easier
      • Ingest events based on how people interact and other data you want to give them
      • shaped.ai
    • Space creation
      • Creating a change object that will be applied to their spaces
    • No use of access policies