Lovely project. Also @rubenvanwyk mentioned SlateDB. I am not sure if this will fit my use-case but, today, I was looking for data hosting options for a self-hosted LLM+bot for email/calendar.
I have this product I have tried and stopped before: https://github.com/pixlie/dwata and I want to restart it. The idea is to create a knowledge graph (use Gliner for NER). Compute would either be on desktop or cloud (instances).
Then store the data on S3 or Cloudflare Workers KV or AWS Dynamo DB and access with cloud functions to hook up to WhatsApp/Telegram bot. I may stick with Dynamo or Cloudflare options eventually though (both have cloud functions support).
I need a persistent storage of key/value data (the graph, maybe embedding) for cloud functions. Completely self-hosted email/calendar bot with LLM, own cloud, own API keys. Super low running cost.
Seems similar ideas, although SlateDB seems a bit more lightweight and using Parquet as primitive (even using Arrow) might mean more compute-heavy on client-side?
Sounds very interesting, but the README has me pondering the downsides. Is the latency very high? Are requests not immediately durable? Is it super expensive?
Lovely project. Also @rubenvanwyk mentioned SlateDB. I am not sure if this will fit my use-case but, today, I was looking for data hosting options for a self-hosted LLM+bot for email/calendar.
I have this product I have tried and stopped before: https://github.com/pixlie/dwata and I want to restart it. The idea is to create a knowledge graph (use Gliner for NER). Compute would either be on desktop or cloud (instances).
Then store the data on S3 or Cloudflare Workers KV or AWS Dynamo DB and access with cloud functions to hook up to WhatsApp/Telegram bot. I may stick with Dynamo or Cloudflare options eventually though (both have cloud functions support).
I need a persistent storage of key/value data (the graph, maybe embedding) for cloud functions. Completely self-hosted email/calendar bot with LLM, own cloud, own API keys. Super low running cost.
SlateDB is awesome, that’s ZeroFS [0] storage backend and it’s been great!
[0] https://github.com/Barre/ZeroFS
How does it compare to https://slatedb.io/ ?
Seems similar ideas, although SlateDB seems a bit more lightweight and using Parquet as primitive (even using Arrow) might mean more compute-heavy on client-side?
Sounds very interesting, but the README has me pondering the downsides. Is the latency very high? Are requests not immediately durable? Is it super expensive?
Yes We'll provide a report to explain how we tradeoff these things, please stay tuned.
How big is the wasm?
License does not yet exist? Hope it’s Apache 2.
For some reason this post links to the dev branch on GitHub, if you switch to the main branch you will see the license file is indeed Apache 2.0.
This is so weird. If you're using this library
1) your serverless and edge runtime needs to have internet access, so it can contact anyone
2) you're obviously not going to be able to efficiently write to S3 while providing guarantees, so it'll be expensive
3) you're writing in rust, so you really care about correctness and efficiency
This seems like a contradiction. Why would you do this as opposed to hosting a redundant postgres on 2 Hetzner/OVH/... servers and writing to that?
Because the means have been given priority over the ends.