How to Clustering cache Keys or Search by multiple Tags in FusionCache?

[fuad-bassam]

Hi, I'm using FusionCache with DistributedCache Ri in a .NET project to cache a large amount of data. I need to retrieve all cache entries that match specific tags efficiently.

Problem
FusionCache allows tagging cached and Removing it RemoveByTag("tag123"), but I couldn't find a method like GetKeysByTagAsync() to retrieve all keys associated with specific tags.

Example scenario:

Cached items have tags Or split (Shard) Cache Keys like Type1:Value2, Type2:Value4, Type3:Value2 and Type4:Value3, I want to retrieve all cache keys that belong to both Type2:Value4 and Type3:Value2.

Questions

1. Does FusionCache support searching for all cache entries associated with tags?
2. If not, what is the recommended way to implement tag-based retrieval?
3. Should I manually maintain an index (e.g., a Redis Set for each tag to store related cache keys)?

Would love to hear any best practices or alternative approaches for handling this in FusionCache! 🚀

[jodydonetti]

Hi @fuad-bassam , in general massive operations like "search by tag" are not really ok on distributed caches, because:

1. they typically don't exist in the related api (g: Redis, Memcached, etc)
2. when using the IDistributedCache abstraction (which is what FusionCache does) there's no such method available

As an example, if I remember correctly even the SCAN command is disabled when Redis is used as a cloud service, for example of Azure Redis Cache, again because it is problematic for performance.

If you are using Redis and really need such a method then yes I would suggest to use something like Redis sets, but again measure everything with real world load and watch out for performance impacts. Also remember that you can have 10 keys expiring at different times and add each of them at different times to the set, so when you'll get the set content back it may contain keys that do not exist anymore, or other problems like this.

Hope this helps!

[fuad-bassam]

Hi @jodydonetti thanks for your time,

What I’m really looking for is a way to search within the cache. From what I’ve found, most caching mechanisms don’t prioritize search or clustering of data, probably because it is not a table-based database, and search in key-value items makes performance challenges, especially when dealing with a large number of cached items.

From your perspective as a specialist, what are the best approaches to handle:

1. Searching or clustering cached items.
2. Retrieving multiple keys in FusionCache (e.g., by prefix, key list, regex pattern, etc.).
3. am I thinking about this issue correctly searching or cluster data conceder that Redis cache isn't like a regular table-based database server, or I need to think about using a hybrid approach using fusionCache with NoSQL database like (Cassandra) for searching

I’d love to hear your perspective on this. Thanks again for your help and for building such an awesome caching library! 🤩

[jodydonetti]

As I said searching in a KV store is not something generally supported, and by going through the IDistributedCache it's not possible at all.

If you want to search in a specific distributed cache, like Redis, I think you should look at what's available in that specific caching service.
For example in Redis there's some support for native tags, see here: by using the Redis-specific commands for tags you should be able to achieve what you want.

For the .NET side of things, I think you should use the StackExchange.Redis package directly, because it is not a "general abstraction for any distributed cache" like IDistributedCache, but a specific client for Redis tht can expore all its features.

Even more so you should look at NRedisStack (built on top of StackExchange.Redis) that adds additional features not currently supported in StackExchange.Redis directly, because they are part of modules or similar. You can see some examples here.

Hope this helps!