EC2 Spot instances, as you probably know, allows clients to use instances on the AWS spot market and is a great fit for people who don’t need computing capacity on a regular basis. It also works for applications that can save their status during check-points in case they get interrupted.
However, not all applications are able to create checkpoints during long hour runs. Mostly these apps run a predictable length of time and are resource-intensive, such as encoding, rendering, modeling, and batch processes.
To support this, AWS EC2 has launched spot instances that can run continuously for a set amount of time of 1 to 6 hours. You can do this through a Spot instance request using the new BlockDuration parameter, which can specify the number of yours you want the instances to run and the maximum price you want to pay. Once that capacity is available on the market for the specified duration, your instances will run for the hourly price. This price is still 30%-45% less than on-demand pricing.
[video_player type=”youtube” width=”560″ height=”315″ align=”center” margin_top=”0″ margin_bottom=”20″ border_size=”3″ border_color=”#46d2c3″]aHR0cHM6Ly93d3cueW91dHViZS5jb20vd2F0Y2g/dj00bWxYWFNpdzk0VQ==[/video_player]