AWS Cost Estimates
Last updated
Last updated
In making the decision to use AWS for PASS production infrastructure, cost will invariably be a deciding factor. This article describes the various cost factors and cost estimates sampling of running such an architecture as described in . Costs will of course be variable and dependent on the exact implementation.
The table below contains the AWS resources used by JHU and the configuration options that will impact the cost for each resource most significantly. The order of the resources is based on percentage of the monthly expense, for instance AWS RDS is the most expensive resource in our implementation. The JHU Configuration column has details for each resource as of September 2024 for what our system currently requires for compute, storage, IO, and monitoring. The can be used to estimate cost for a PASS application AWS architecture to your specifications.
Cost Factors
Database Instance Class
Storage Size/IO Bandwidth
Snapshot retention
Data Transfer
Database License
JHU Configuration
AWS RDS PostgreSQL
Multi-AZ
Instance Class: db.m5.large
Storage: SSD (gp3) 500 GiB
Snapshot retention: Daily system snapshot 7 days (Default)
Cost Factors
ECS Task vCPU/Memory allocation
Ephemeral Storage > 20GB
JHU Configuration
3 ECS Tasks run for deposit and notification services 24/7
Per task 1vCPU/4GBMem
4 ECS tasks run for data loader jobs for 2 hours/day
Per task 2vCPU/4GBMem
20GB ephemeral storage for each Task
Cost Factors
EC2 Instance Type
Storage Size/IO Bandwidth
Data Transfer
JHU Configuration
1 EC2 for Pass-core/Pass-UI, Instance Type: m5.large, Storage: EBS 40GiB
1 EC2 for jumpbox, Instance Type: t3.micro, Storage: EBS 8GiB
Cost Factors
Size of Logs
Number of Custom Metrics
Number of Dashboards
JHU Configuration
7 alarms
3 custom metrics
1 Dashboard
1 Canary Script
Cost Factors
Usage rate
Data transfer
JHU Configuration
2 ALBs
3 Listeners and Rules
Cost Factors
Number/type of Rules
Bot Control
JHU Configuration
3 Rules
1 Bot Control
Cost Factors
Storage Type
Storage Size
JHU Configuration
~ 5 Buckets
~ 3 GB Total as of Sept 2024
JHU’s PASS production architecture detailed above results in a fairly-consistent operating cost totaling $700-$800 per month.
About half of the spend is dedicated to the High Availability (HA) production database.
Compute resources comprise roughly a quarter of the spend.
The remainder of the cost is spread across supporting services, such as monitoring logging, and security.
Storage costs for PASS are currently negligible.
RDS
$379
ECS
$134
EC2
$76
CloudWatch
$50
EC2-Other
$39
Elastic Load Balancing
$33
WAF
$15
VPC
$14
CloudTrail
$12
Config
$3.82
DevOps Guru
$3.12
S3
$1.54
Lambda
$1.04
SQS
$0.24
SSM
$0.19
ECR
$0.18
Route 53
$0.10
Key Management Service
$0.06
Total costs
$772