|
|
|
|
|
|
by ryanburk
3557 days ago
|
|
|
re: scale and on demand: "you only pay for the compute that you need, down to a 100ms boundary in the case of AWS Lambda" it will be interesting to see how small these boundaries can become before it is too expensive to measure at the level of granularity sought. when you can spin up a container, run job, and destroy faster than you can even log. |
|
|
Either you pay to idle 1 GB of ram on some giant server somewhere, or you put it on really fast SSD and pay the time to load it all into ram.
Now a little Python or C routine that uses 8 MB of ram... becomes a much easier thing to do.