correct, which is why

LookupResources

exists

however, for very large datasets, neither one will scale, which is why https://github.com/authzed/spicedb/issues/207 is being worked on

This is helpful, thanks !

the first pieces for 207 are in review now

Will it be available in SpiceDB or AuthZed enterprise only?

both, but for SpiceDB if you intend to use the Tiger cache, you'll need to run it yourself

Authzed will have support for it built in

Hi Joey thanks, I have added this issue https://github.com/authzed/spicedb/issues/613

thanks!

Is there any limit on no of namespaces in Schema? We have more 40+ micro-products in our enterprise , to provide isolation, want to use namespace for each microproduct.

the only limit right now is the size of the schema

in bytes

How much is that?

256KiB currently

but we're likely to raise it

Thanks this helps, to run the spiceDB in production with ~ 1 million permissions checks daily, not much nested permissions, what could be recommended hardware, even a very high level estimate or reference would be helpful.

Hi all, I noticed that there was a scheduled maintenance on May 13th. As I can see from API availability and Response Duration metrics, it had not effect the system at all. May I ask if the scheduled maintenance window was excluded in calculation of these metrics?

It wasn’t excluded, there just wasn’t any measurable downtime. The maintenance was to upgrade cockroach db.

That should run on pretty much anything. You will get better performance running on boxes with more CPUs, so maybe something with like 4 vCPU and 8gb of memory.

3 instances, and each instance having 4 vCPU and 8 gb memory should be good enough?

That should be overkill

Ok, as I am planning to run it inside Kubernetes cluster, what you recommend?

I mean, that’s a good proposal, it should take you far beyond 1m requests per day.

i'm looking at the python API, and I'm seeing that the client sets up a

secure_channel

by default. if I want to use that locally with a non-TLS server, i take I'll have to write my own client class that uses the insecure channel?

use `insecure_bearer_token_credentials`: https://github.com/authzed/authzed-py/blob/main/tests/v1_test.py#L29

ah, thank you!

looking at the WatchService: does that emit events when a write is received by SpiceDB, or does it emit an event when the write lands?

after it lands and is committed

it reflects changes that have been applied