3 ms·
100% agree, we migrated from Redis Labs -> Elasticache and it was cheaper, more performant, and closer to the rest of our infrastructure. We lost "support", but
by drchopchop 7y ago
100% agree, we migrated from Redis Labs -> Elasticache and it was cheaper, more performant, and closer to the rest of our infrastructure. We lost "support", but the service is so brain-dead simple that I can't imagine what the need for that would really be. I wish it auto-scaled like Dynamo/Aurora, however it's just a few clicks to up the instance size, so not a super-pressing need.