Hi All,
We’ve been using Redis as the distributed cache and would want to optimize the cost of Redis using DRAM & the costly network call for each request.
We’ve been looking at the embedded databases which runs on SSDs and provide DRAM like performance & avoid costly network calls from the application. This kind of decentralized architecture certainly requires some smarts in the application.
Would badger be an ideal choice for such use cases ? How does full on-disk Badger perform against Redis / Aerospike ?
I’m using the latest version of badger this way
db, err := bv3.Open(bv3.DefaultOptions("/tmp/badger"))
if err != nil {
panic(err)
}
and loading 10M keys. Key is 0-10M integer string and value is 25bytes long. I believe badger persists on disk async but the RAM usage of the idle golang application is dancing b/w 300 MB - 600 MB. Why’s the memory so high ? is it not persisting the data to disk & instead storing data in DRAM ? Or am I missing any other setting in Badger?
Regards