How to build a JavaScript Policy pulling data from a local data storage

[anton]

Hello,

I'm looking for an expert advice please.

I need to build an API call that does this:

1. Receives a string ID as query parameter
   
2. Looks up a record in the Apigee data storage identified by the given ID string. The record is super simple - only two fields - ID and Data. Data is up to 4096 bytes long.
3. If a record found, the code calls some target endpoint passing on the found data and returns the response.
4. If not found, it calls a different target endpoint that generates the Data, and the code creates a new record in the data storage, and then does #3.

I would greatly appreciate some simple code please as well as general advice on the best approach.

Lots of thanks in advance,

Anton.

[ssvaidyanathan · Accepted solution]

HI @Anton+Bar

Please find the attached zip. See if that works.

Assumptions I made to build the proxy: I am pointing it to httpbin.org/get end point. I wrote another mock service that generated a Guid when the data is not available in Cache.

Created an Env cache called "demoCache"

Hope this gives you some idea

cachingtest.zip

[anton]

Thank you Sai! Quick question - is the caching persistent or will it "forget" the entries after some time? I was researching after I posted this and bumped into KeyValueMap but couldn't make it work so far.

[ssvaidyanathan]

Yes, its time based and transient. Thats why I recommend that. KVM is not for that purpose. For more info on Caching, check this doc

[ssvaidyanathan]

@Anton+Bar - please accept this answer if it resolved your issue

[anton]

I will accept when it will. Sorry, still not with you. I'm looking for persistent data storage, not something like cache.

[ssvaidyanathan]

Sure. My question to you is - you have a source of truth anyways, the callout which provides that info, then why does this need to be persisted always ? We can always go back to the backend system which I guess is the source of truth.

[anton]

Yes, that's true, but to get that data string might take a relatively long time (long story as to why it is so), so the first call can take around 30 seconds. I'd like to avoid that.

[ssvaidyanathan]

Ok. You can have a longer Cache TTL in that case ? What if the values in the source change, how will that update KVM ? In other words, you will need to sync both the data sources. In stead have one as primary and the other a temp data store. Again this is my 2 cents.

[anton]

Good point! However in our case the values in the source do not change, so I'd be happier with KVM. Would that work?

[ssvaidyanathan]

Yes it can. How many records do you think will be stored ? The reason I tell this is its generally recommended to store config in there. However its a persistent store.

[anton]

I see, so we'll probably use Cache then.

[ssvaidyanathan]

Awesome !!!