As an alternative of 1 occasion per VM, you’re now in a position to stack a number of cases behind a Redis proxy. There’s one other large change: Though you continue to use two nodes, each nodes run a mixture of main and reproduction processes. A main occasion makes use of extra assets than a reproduction, so this strategy helps you to get the very best efficiency out of your VMs. On the identical time, this mixture of main and reproduction nodes mechanically clusters information to hurry up entry and allow help for geo-replication throughout areas.
Azure Managed Redis has two completely different clustering insurance policies, OSS and Enterprise. The OSS possibility is similar as utilized by the group version, with direct connections to particular person shards. This works effectively, with close-to-linear scaling, nevertheless it does require particular help in any shopper libraries you’re utilizing in your code. The choice, Enterprise, works via a single proxy node, simplifying connection necessities for purchasers on the expense of efficiency.
Why would you employ Redis in an software? In lots of instances it’s a instrument for holding usually accessed information cached in reminiscence, permitting fast learn/write entry. It’s getting used anyplace you want a quick key/worth retailer with help for contemporary options akin to vector indexing. Utilizing Redis as an in-memory vector index helps hold latency to a minimal in AI purposes based mostly on retrieval-augmented era (RAG). Cloud-native purposes can use Redis as a session retailer to handle state throughout container purposes, so AI purposes can use Redis as a cache for current output, utilizing it as semantic reminiscence in frameworks like Semantic Kernel.
