AWS Fargate may actually be good for your (company’s) wallet.

When AWS Fargate launched, there was a wide consensus that the pricing was too high, with the move seen as AWS choosing to target companies wanting to run container workloads without Kubernetes. However, with the advent of AWS EKS backed by Fargate Profiles, and ECS with Fargate Spot compatibility, you may wish to consider this product once more.

Since nobody else has made a basic pricing table for my region (ap-northeast-1, Tokyo), I’ll share this with you below. (No guarantee on the pricing being correct now, or in the future though!)

For workloads, it may be worth considering Fargate Spot for things such as CI builds, extra scaling above base load and asynchronous data transformation. Normal Fargate for critical workloads such as non-idempotent processing of data. With a guarantee of 99.99% uptime, this may be an excellent option for your next compute project.

But why should I use Fargate?

As of right now, the most common way to provide Compute on AWS is via EC2 instances. However, this comes with a tradeoff between you and AWS, which is that you are entirely responsible for:

  • capacity for autoscaling
  • optimisation of your instance types (choosing the right size)
  • orchestration technology (including attempting to reduce overhead)
  • security, networking, and maintenance

But with Fargate, you do not have to worry about most of these things.

  • Fargate has virtually unlimited capacity (depending on account limits and whether you choose Fargate Spot).
  • No underlying nodes to worry about, you choose a CPU/Memory combo and that’s what you get served.
  • You can use Fargate with EKS for the benefit of Kubernetes, and never run node groups in EC2, meaning your base load can be 0, resulting in a further reduction of costs.
  • Security, networking and maintenance are all handled by AWS. For instance, during the Spectre/Meltdown incidents, AWS patched Fargate, but anyone running shared workloads on EC2 was left to orchestrate and perform a patch to their fleets manually.
  • As mentioned previously, AWS guarantees Fargate availability to 99.99%.
  • The massive amount of engineer time saved by using a solution like Fargate cannot be measured. That’s up to how you or your organisation works.

Pricing Chart (for Tokyo)


ap-northeast-1
CPU/hr Memory/hr Total per 1hr build Total Monthly Spot CPU/hr Spot Memory/hr Spot Total per 1hr build Spot Total Monthly
0.25CPU/0.5GB 0.01264 0.002765 0.015405 11.24565 0.0038293375 0.00083767 0.0046670075 3.406915475
0.25CPU/1GB 0.01264 0.00553 0.01817 13.2641 0.0038293375 0.00167534 0.0055046775 4.018414575
0.25CPU/2GB 0.01264 0.01106 0.0237 17.301 0.0038293375 0.00335068 0.0071800175 5.241412775
0.5CPU/1GB 0.02528 0.00553 0.03081 22.4913 0.007658675 0.00167534 0.009334015 6.81383095
0.5CPU/2GB 0.02528 0.01106 0.03634 26.5282 0.007658675 0.00335068 0.011009355 8.03682915
0.5CPU/3GB 0.02528 0.01659 0.04187 30.5651 0.007658675 0.00502602 0.012684695 9.25982735
0.5CPU/4GB 0.02528 0.02212 0.0474 34.602 0.007658675 0.00670136 0.014360035 10.48282555
1CPU/2GB 0.05056 0.01106 0.06162 44.9826 0.01531735 0.00335068 0.01866803 13.6276619
1CPU/3GB 0.05056 0.01659 0.06715 49.0195 0.01531735 0.00502602 0.02034337 14.8506601
1CPU/4GB 0.05056 0.02212 0.07268 53.0564 0.01531735 0.00670136 0.02201871 16.0736583
1CPU/5GB 0.05056 0.02765 0.07821 57.0933 0.01531735 0.0083767 0.02369405 17.2966565
1CPU/6GB 0.05056 0.03318 0.08374 61.1302 0.01531735 0.01005204 0.02536939 18.5196547
1CPU/7GB 0.05056 0.03871 0.08927 65.1671 0.01531735 0.01172738 0.02704473 19.7426529
1CPU/8GB 0.05056 0.04424 0.0948 69.204 0.01531735 0.01340272 0.02872007 20.9656511
2CPU/4GB 0.10112 0.02212 0.12324 89.9652 0.0306347 0.00670136 0.03733606 27.2553238
2CPU/5GB 0.10112 0.02765 0.12877 94.0021 0.0306347 0.0083767 0.0390114 28.478322
2CPU/6GB 0.10112 0.03318 0.1343 98.039 0.0306347 0.01005204 0.04068674 29.7013202
2CPU/7GB 0.10112 0.03871 0.13983 102.0759 0.0306347 0.01172738 0.04236208 30.9243184
2CPU/8GB 0.10112 0.04424 0.14536 106.1128 0.0306347 0.01340272 0.04403742 32.1473166
2CPU/9GB 0.10112 0.04977 0.15089 110.1497 0.0306347 0.01507806 0.04571276 33.3703148
2CPU/10GB 0.10112 0.0553 0.15642 114.1866 0.0306347 0.0167534 0.0473881 34.593313
2CPU/11GB 0.10112 0.06083 0.16195 118.2235 0.0306347 0.01842874 0.04906344 35.8163112
2CPU/12GB 0.10112 0.06636 0.16748 122.2604 0.0306347 0.02010408 0.05073878 37.0393094
2CPU/13GB 0.10112 0.07189 0.17301 126.2973 0.0306347 0.02177942 0.05241412 38.2623076
2CPU/14GB 0.10112 0.07742 0.17854 130.3342 0.0306347 0.02345476 0.05408946 39.4853058
2CPU/15GB 0.10112 0.08295 0.18407 134.3711 0.0306347 0.0251301 0.0557648 40.708304
2CPU/16GB 0.10112 0.08848 0.1896 138.408 0.0306347 0.02680544 0.05744014 41.9313022
4CPU/8GB 0.20224 0.04424 0.24648 179.9304 0.0612694 0.01340272 0.07467212 54.5106476
This is a sample ap-northeast-1 Fargate pricing table.
Prices will not be updated, but should be current as of July 19th, 2020.
No guarantees on data accuracy or effectiveness.

Published by Alexander

- Alexander is a professional Operations (DevOps/NetOps/SysOps) SRE and Developer living in Tokyo.