HybridCache - tags and invalidation about aspnetcore

[mgravell]

@jodydonetti you might be interested in this; am I on entirely the wrong path here? would any of this be useful to you? is any of it "that's so close to being useful, but not quite" ?

[mgravell]

@sebastienros might also be of interest to you

[jodydonetti]

Hi @mgravell , I'll look into this later today, can't wait to see the details of this 😬
Will update asap.

[verdie-g]

I understand this proposal is to receive invalidations from the distributed cache layer (aka L2) and would not support invalidations through Kafka or "SQL polling". That's an available feature on Redis but is that a common feature from other NoSQL database systems? With the HybridCache proposal (#53255), what exactly would be the blocker to implement a Redis channel-based invalidation mechanism?

[mgravell]

@verdie-g the entire point is to provide an abstraction upon which any L2 could provide invalidations, if it is so capable. The same thing would be possible in SQL polling by adding a couple of tables - one for the tag last-known expiration state, and one for some backlog of invalidations (the latter could be fed via a trigger or explicit code), and having a periodic timer that simply checks for new data (using an incrementing counter) - that could be every N seconds, configurable. There may also be some SQL mechanism involving SQL service-broker queues, but I don't think (from memory) that this has the correct multi-receiver semantics.

I'm not a Kafka expert (I've simply never had cause to dabble), but assuming that Kafka queues support a mode that allows multiple recipients to receive the same message, such a queue could absolutely be used as the backbone for invalidation for an L2 backend that wanted to do so. For comparison, the same intent could also probably be described by Redis "streams", which are, IIRC, broadly similar to Kafka queues.

With the HybridCache proposal (#53255), what exactly would be the blocker to implement a Redis channel-based invalidation mechanism?

Not fully sure I understand the question, but the biggest blocker is the lack of an abstraction API that allows the L2 backend (currently described by IDistributedCache) to communicate the intent of a channel-based invalidation mechanism that HybridCache can then consume. Which is exactly the gap that this proposal is trying to plug ;p

[dotnet-policy-service]

Thank you for submitting this for API review. This will be reviewed by @dotnet/aspnet-api-review at the next meeting of the ASP.NET Core API Review group. Please ensure you take a look at the API review process documentation and ensure that:

• The PR contains changes to the reference-assembly that describe the API change. Or, you have included a snippet of reference-assembly-style code that illustrates the API change.
• The PR describes the impact to users, both positive (useful new APIs) and negative (breaking changes).
• Someone is assigned to "champion" this change in the meeting, and they understand the impact and design of the change.