2 ms·
Read performance is much easier to scale (in one box or several) than write performance. It's usually the writes that make you look at Cassandra and similar, in
by nine_k 6y ago
Read performance is much easier to scale (in one box or several) than write performance. It's usually the writes that make you look at Cassandra and similar, instead of adding more disks and RAM, or spinning another read-only replica.
24 NVMEs should have a lot of write throughput, though.