Hi Growthbook team,
Is there any way to set up attributes on growthbook such that if we pass any parameter to growthbook, it analyzes that under which experiment that parameter falls by using the attributes that we have set.
lets say we have provided growthbook access to our database and provided it parameters in the database which it should analyze.
Now let's say there's an api endpoint and we need to know which experiment that url falls into based on the parameters we have asked growthbook to analyze.
So to summarize what I want to convey is that instead of sending attributes from our backend service and get to know which experiment it falls under.. can we configure attributes on growthbook side by providing it access to our database and growthbook then helps tell us which experiment any parameter(url for example) falls under
11/22/2022, 2:06 PM
We don't support this currently. You could build your own back-end service though. It could use one of our back-end SDKs, connect to your database to populate attributes, and expose an API endpoint to get a variation.
We're planning to build a proxy server similar to this in the future.
12/21/2022, 3:57 PM
@future-teacher-7046 could you explain more about this proxy server? Which use cases does this cover? Is it meant as reverse proxy to tunnel request to a frontend through a GB instance?
12/21/2022, 4:34 PM
Initially it will serve 2 primary purposes.
1. Highly scalable cache/proxy for the GrowthBook API that can support millions of client SDK connections.
2. Option to shift feature flag evaluation to the proxy server so you don't expose all of your rule evaluation logic to client SDKs. Instead, you just get back the state of all the features for a specific user.
12/22/2022, 2:06 PM
I see ... but does the 2nd point fit to your general API based approach? Or do you aim to move SDK logic to the API?
12/22/2022, 2:20 PM
We want to support both. Evaluating feature flags within the SDK is ideal for back end applications and when you want full control over how things are implemented. Evaluating feature flags in an API call is great for client-side apps when you don't want to expose all of the rules to users or if you prefer a simple http request instead of an SDK.