Arun Sirpal continues a series on Azure Redis:
Remember – basic should never be used for production. Also, if you need dedicated service then you will not want C0 because this is based on shared infrastructure. Redis can get expensive but could be cost – effective especially if you design to use a multi app approach per cache.
I select P1 – Premium with 6GB cache just to talk a couple things through.
As a note, 6GB of cache is a lot in most environments. That’s because your average cached element size in Redis should be measured in single-digit or double-digit bytes, not kilobytes. You’re typically caching individual values, not entire documents, so if you average 64 bytes per cached key-value combo, you can get somewhere around 90 million values in cache at a time. The database call savings add up quickly, considering a really simplistic estimation: if the average number of queries before expiration for a cached item is 3, a single “cycle” of caching saves you about 270 million database calls. That can allow you to downscale your relational databases considerably, saving a lot of money in the process. There’s a lot of hand-waving I’m doing in the math and a lot of complexity I’m wiping away, but both of those tend on average to make the cache more effective, not less.