- 最新
- 投票最多
- 评论最多
Yes I think you missed the points of the difference between the two.
Provisioned means you have a max number of writes/reads per second your table can do, you do not pay per-read, you pay per hour, this is ideal if your table is going to have a predictable traffic, it gets better with the auto scaling, so let's say you have double the traffic during the night, that still is a predictable scenario and provisioned with auto scaling is the best solution.
Now the on-demand is when you have no idea how your traffic is going to come, maybe you get a sudden surge of 1 million requests in one hour but no requests on the rest of the day, provisioning for this would end up too expensive or unreliable as it could throttle.
Remember that provisioned will throttle if you don't have enough RCUs, and your request will fail.
The provisioned throughput costs are for reads or writes per second for one hour. So given your example of $0.0001544 per read, if you use all of your provisioned throughput you can do 3600 reads at one per second over an hour, so the per read price is $0.000000042888889 ($0.0001544/3600), or about $0.043 per million reads.
相关内容
- AWS 官方已更新 1 年前
- AWS 官方已更新 1 年前