We ran a little experiment building git on top of Tigris. Check out this post from Xe on the learnings
Tigris Data
736 posts
Tigris Data
@TigrisData
Tigris is a globally distributed S3-compatible object storage service that provides low latency anywhere in the world. Now with bucket forking.
Every git repo you push is really just a pile of objects: your files, compressed and content-addressed, plus a few tiny pointers that move around. That's the whole storage model.Replying to @TigrisDataPushing a big repo became a stat() storm with hundreds of thousands of round trips, most of them answered "404, nothing here." Git might be the best filesystem-latency benchmark we've ever run, entirely by accident. If your storage has a weak spot, git will find it.The full writeup (with every disaster and the fixes) is here:- Every time a user downloads something from your app, you get charged. With Tigris Data, your bill stays the same whether you have 10 users or 10,000. We are super happy to see Rishi Raj Jain and LaunchFast support Tigris!Our egress-free business model is important to keeping margins low and remaining competitive as a growing SaaS. Tigris is truly global Sydney loads as fast as Toronto - you don't have to set up a CDN. Thank you for including us Rishi!
- Fifty agents, one bucket. They can't share a bucket without overwriting each other, and copying it fifty times is slow and expensive. StorageSDK gives object storage git semantics. Snapshot a bucket, then fork it. A fork is copy-on-write: constant time, shares the parent's
GIFForking a 50GB dataset fifty times doesn't cost 50x storage. You only pay for what each fork changes. Each agent writes into its own fork. The parent and every sibling stay untouched. Read More: - Every object store has its own CLI. AWS, wrangler, gcloud, az. @TigrisData and @computesdk built one CLI for all of them: github.com/storagesdk/sto… List a bucket, switch providers by changing one word. It pipes like a real Unix tool. It can snapshot and fork your data (git
GIF - Tigris Data repostedContinuing with the Tigris feature releases, here is the latest one: Bundle API that allows you to batch download files instead of sending thousands of individual requests.
- Tigris Data reposted
Replying to @michelleefangTuesday 6/16 (cont'd) ‣ Big Data 3.0- Data Systems for Large Scale Research luma.com/2z8lp24e @SpiralDB @TigrisData @krea_ai ‣ Codex Community Meetup - San Francisco luma.com/codex-jun16-20… @itsajchan ‣ Genloop Retail Analytics Meetup @ Databricks Summit - Trying to do AI training on your dataset but you created everything in it as individual objects? Wouldn't it be great if those freakin' storage nerds would just give you a zipfile with everything you want in it?Tigris does that with the bundle API! Fetch a bundle of objects and go off to the races making the world a better place one SaaS app at a time!
- Agents sometimes delete the wrong thing. That's no longer a problem: tigrisdata.com/blog/soft-dele… We just added Soft Delete to Tigris. You can flip it on once, pick a retention window, and every delete just becomes recoverable.
GIFAn object, a prefix, even a whole bucket someone dropped by accident comes back. Your agents can move fast and delete freely, and nothing's actually gone until you say so. Take a look: - Tigris Data repostedTigris is the only storage service that offers git semantics for storage and we are continuously making it betterS3, R2, GCS, Azure all great storage options - but can they fork buckets? Turns out storage for agents is different, and our compute-based friends need to experiment and drop buckets at will. storagesdk fixes that with one idea: snapshots and forks as first-class operations.
GIF - S3, R2, GCS, Azure all great storage options - but can they fork buckets? Turns out storage for agents is different, and our compute-based friends need to experiment and drop buckets at will. storagesdk fixes that with one idea: snapshots and forks as first-class operations.GIF
