Cost Estimate

This document provides a monthly AWS cost estimate for deploying the MOCM platform using the OpenTofu configurations in terraform/aws/. All figures are based on ap-northeast-1 (Tokyo) On-Demand rates as of early 2026, assuming 730 hours/month. Prices can change—always verify with the AWS Pricing Calculator when modifying region or resource types. These estimates exclude outbound data transfer beyond ~50GB, AWS Support, Shield Advanced, and any service/component not provisioned by the stack (e.g. CloudFront). WAF costs are in Section 6.

1. Deployment Scenarios - Quick Comparison

This stack supports two production-level deployment scenarios. Choosing between them depends on cost sensitivity and uptime requirements.

Notes:

• Estimates include compute, managed services, networking, storage.
• Excludes outbound data over ~50 GB, premium support, and services not managed by this stack.
• WAF is included by default (see Section 6).

How to select: Copy the relevant .tfvars file to terraform.tfvars before running OpenTofu.

xxxxxxxxxx

# Minimum (cost-optimized)

cp terraform.tfvars.cost-optimized.example terraform.tfvars

# Production (high availability)

cp terraform.tfvars.high-availability-multi-az.example terraform.tfvars

2. Scenario A — Cost-Optimized (Minimum)

Config file: terraform.tfvars.cost-optimized.example

This profile is for users minimizing AWS spend for production loads. It uses a single NAT, SPOT nodes for EKS, and a single RabbitMQ broker. If an AZ fails, network egress is lost until recovery. SPOT nodes may be interrupted.

SPOT node note: For EKS, only t3.large, t3.xlarge, and t3.2xlarge are allowed. Costs vary with which size and discount rates the autoscaler obtains.

Key Terraform settings:

xxxxxxxxxx

single_nat_gateway             = true

eks_node_group_capacity_type   = "SPOT"

eks_node_group_instance_types  = ["t3.large", "t3.xlarge", "t3.2xlarge"]

eks_node_group_desired_size    = 10

eks_node_group_min_size        = 10

rabbitmq_deployment_mode       = "SINGLE_INSTANCE"

elasticache_node_type          = "cache.t3.medium"

mongodb_instance_type          = "t3.large"

enable_bastion                 = false

Monthly Service Cost Breakdown:

3. Scenario B — High Availability (Production)

Config file: terraform.tfvars.high-availability-multi-az.example

For environments demanding maximum up-time, this mode provisions multi-AZ networking, On-Demand EKS nodes, and RabbitMQ in a 3-node cluster to eliminate single points of failure.

Key Terraform settings (diffs from Cost-Optimized):

xxxxxxxxxx

single_nat_gateway             = false

eks_node_group_capacity_type   = "ON_DEMAND"

eks_node_group_instance_types  = ["m5.xlarge"]

eks_node_group_desired_size    = 8

eks_node_group_min_size        = 8

rabbitmq_deployment_mode       = "CLUSTER_MULTI_AZ"

elasticache_node_type          = "cache.t3.medium"

mongodb_instance_type          = "t3.xlarge"

enable_bastion                 = true

bastion_instance_type          = "t3.medium"

Monthly Service Cost Breakdown:

4. Key Cost Drivers — How Variables Affect Your Bill

The variables/choices below most significantly affect monthly AWS charges. Adjust these in .tfvars to scale usage versus cost.

TDB

5. AWS WAF

WAF is included in the base stack, with 3 managed rules active by default.

Baseline cost (default rules):

Advanced managed rules (disabled by default):

Shield Advanced ($3,000/mo, not included) is opt-in outside the stack.

WAF Logging: If enabled, logs go to S3 at ~$0.023/GB/month stored.

6. Get a Precise Quote — Calculator & Planning Steps

To precisely estimate AWS billing:

1. Run tofu plan to get actual resource sizes/counts.

2. Use the AWS Pricing Calculator:

   • Region: ap-northeast-1
   • EC2 details (EKS + MongoDB + Bastion)
   • Amazon MQ/RabbitMQ config
   • ElastiCache sizing
   • Count of NAT Gateways & egress
   • Count of VPC endpoints × AZs
   • EBS volumes
   • WAF config (ACL, rules, expected traffic)

3. Add a 10–15% buffer for data transfer, logging, and S3 request cost.

4. New accounts: review AWS Free Tier.

7. Observability Stack (Optional Add-On)

(Only applies if the monitoring node group is enabled. See OBSERVABILITY_GUIDE.md for setup details. Pricing uses us-east-1 as reference; adjust for your region.)

Compute: EKS Monitoring Node Group

Dedicated Spot node group. ResourceSpecPrice/hrCountMonthlyt3.xlarge4 vCPU / 16 GiB$0.0502–3$73–$110t3.2xlarge8 vCPU / 32 GiB$0.1002–3$146–$219

Default config: t3.xlarge, min 2, desired 3, max 5.

Storage: EBS (gp3)

ComponentSize/ReplicaReplicasTotal$/movmstorage50 GiB3150$12vlstorage100 GiB3300$24Root EBS50 GiB3150$12Total$48

Network: Internal NLB and Cross-AZ Data

Deep Dive: NLB Cost Model

• NLB hourly charge: $0.0225 per AZ × 3 AZ × 730h = $49.28/mo
• NLCU usage (for this workload): Negligible (<$1/mo)
• Cross-AZ transfer (MongoDB log flows, etc): ~$0.80/mo

See “Cost Optimization” below for details and ways to reduce/remove NLB cost.

Total: CategoryLowHighCompute (Spot)$73$110Storage (gp3)$48$48Network (NLB 3-AZ + cross-AZ)$51$51Overall Total $172 $209

Managed Alternatives Comparison

SolutionEst. CostNotesSelf-hosted (3-AZ NLB)$172–$209Full control; no per-metric/log feesSelf-hosted (1-AZ NLB)$139–$176One AZ; lower availability for logsNo NLB (K8s logs only)$122–$159No MongoDB EC2 log shippingAmazon CloudWatch$300–$800+Cost scales by metric/logAmazon Managed Prometheus + Grafana$200–$500+Per-series/metric queries chargeDatadog$500–$2,000+Per-host pricing + log indexing fees

8. Cost Optimization Tips

Apply these tips to minimize your AWS cost for both cluster and observability stack:

Core Stack

1. Keep single_nat_gateway = true unless you require multi-AZ NAT. Saves ~$70/mo.
2. Use SPOT nodes (eks_node_group_capacity_type = "SPOT") for the highest savings.
3. Start RabbitMQ with SINGLE_INSTANCE and increase HA only when needed.
4. Keep log retention periods minimal (e.g., 3 days for MQ/EKS) unless compliance requires more.
5. Minimize VPC Flow Logs retention (defaults: 7 days).
6. Use ECR lifecycle policies to limit docker image sprawl/storage.
7. Default S3 encryption (AES256) avoids extra KMS expense.

Observability / Monitoring

1. Right-size storage: Start with smaller PVCs and expand as you grow.
2. Reduce replicas in non-prod (use single-node clusters).
3. Shorten log/metric retention to minimize storage growth.
4. Handle Spot interruptions: If Spot instance churn is too frequent, fallback to On-Demand (will 3× compute cost).
5. Remove unused Grafana dashboards to avoid unnecessary fetches.
6. Skip NLB if MongoDB log shipping is not required—use ClusterIP for VLInsert service to save ~$51/mo.
7. Limit NLB to 1 AZ for log shipping to save ~$33/mo, if you accept the single-AZ failure risk.

/table