radiofmd

Still way to slow (coldstart, throughput and general latency), I was referring to the latest version which has still a convoluted, nonsense stack. I'd stay away from Prisma, there's enough info online.

I am pretty excited about D1 and in general a CF fanboy, so congrats to this huge milestione. But, we need more--following is just a wish list:

- Development since start took time (with a good reason), hope we see soon read replication and higher pace

- So, some roadmap _with_ ETAs would be great, e.g., will replication come this or next year?

- I don't care about read performance because I know that sqlite paired with CF's infrastructur will perform but I would love to have more infos on write performance; this is really sqlite's weakest point and I haven't seen any implementation which could compete anywhere with other dbs; sqlite slows down very quickly when doing a couple of small writes/sec

Then, I'd like to see the close competition here on HN, so db-providers with many read-replicas (CEO's and/or devs from Turso and Fly/LiteFS), commenting on D1 and how they compare against and what they plan to compete. This is a too exciting space and time to be laissez-faire.

Great link & insights from Kenton which shows once again that CF is on the edge (what a pun haha).

> standard ORMs like Prisma

Prisma isn't standard and too slow anyway.

This. If we wanted opaque pricing with egress, compute and end-of-month-surprises I'd go with AWS or Google Cloud.

Nah, D1 is in a different league. One key feature of D1 is having free read replicas around the globe. That's CF's DNA and I expect this to happen this or latest Q1/24, and again, for free (so, without any surcharge). Re storage: You get ofc more than 1GB on the paid plan.

Neon doesn't have read replicas and even if they already work on it, I wouldn't expect it before 2025 if at all and never at CF's pricing (Neon still charges for egress).