Many customers are not using Spots because their workloads aren't built to support unexpected interruptions.
Spot blocks provide up to 6 hours of uptime.
It will allow our customers to run short stateful workloads, such as Jenkins tasks, automated tests, map-reduce, batch, ML training, Video transcoding, etc - on spots.
For example, A customer is transcoding hundreds of videos every day, and the process takes hours with standard instances.
He wants to accelerate the process, and his options are:
Allocate larger instances > will increase the OPEX
Develop support in Spots > this could be tricky and increase the R&D cost
A possible solution with Spot0:
Combine duration-defined spots with On-demand fallback to offer greater stability for short workloads that cannot tolerate interruptions.