AWS vs. GCP Costs

 

In this post we will review price comparison of a kubernetes deployment on AWS and GCP. 

We will check a deployment that :

• Includes 10 nodes of 8 CPUs and 16GB RAM
• Serving 1000 requests per second 
• The size of the request and the size of response is 1KB

General Calculations

The egress traffic bandwidth per month is:

1000 RPS * 86400 seconds in a day * 30 days * 1 KB per response = 2592000 GB

GCP Monthly Costs

Kubernetes cluster up-time:

0.1$ per hour = 72$ per month

Kubernetes cluster nodes:

10 nodes of 8 CPUs 16GB machine (c2d-highcpu-8) = 10 * 137$ = 1370$

Egress:

2592000 GB * 0.01 per GB = 25920$

Load balancer up-time:

0.025 per hour = 18$

Load balancer inbound traffic cost:

0.008$ per GB * 2592000 GB = 20736$

Load balancer outbound traffic cost:

0.008$ per GB * 2592000 GB = 20736$

Total monthly cost: 72834$ 

AWS Monthly Costs

Kubernetes cluster up-time:

0.1$ per hour = 73$ per month

Kubernetes cluster nodes:

10 nodes of 8 CPUs 16GB machine (c2d-highcpu-8) = 10 * 149$ = 1490$

Egress:

2592000 GB * 0.09 per GB = 233280$

Load balancer up-time:

0.0225 per hour = 16$

Load balancer outbound traffic cost:

0.00864$ per GB * 2592000 GB = 22394$

Total monthly cost: 257253$ 

Final Note

We can see that the pricing of the cloud providers is similar differs most regarding the traffic ingress and egress. The major cost depends on the traffic, and not on the compute resources, as the traffic cost is high, and it also affects the high costs of the load balancer.