Remove unnecessary AWS resources, VPC + NAT gateway - amazon-web-services

Question

I recently set up an application on AWS via CDK. The application consists of a Dockerized nodejs application, which connects to an RDS instance, and has a Redis caching layer as well. After having the application deployed for a few days, the costs are much higher than I had anticipated, even with minimal traffic. After looking through the cost explorer, it looks like half of the cost is coming from the NAT gateways.
In my current setup, I have created two VPCs. One is used for the application stack, and the other is for the CodePipeline. I needed to add one for the pipeline because without it I was hitting rate limits when trying to pull Docker images during the CodeBuildAction steps.
I'm not very comfortable with the networking bits, but I feel like there are extra resources involved. The pipeline VPC has three NAT gateways and three EIPs. These end up just sitting there waiting for the next deployment, which seems like a huge waste. It seems like a new gateway + EIP is allocated for each construct the VPC is attached to in CDK. Can I just make it reuse the same one? Is there an alternative to adding a VPC at all and not getting rate limited by Docker?
I also find it very surprising (I might just be naive) that the NAT gateway is so far equally as expensive as my current Fargate task costs. Is there an alternative that would serve my purposes, but come at a little lower cost?
Anyways, here are my two stacks:
// pipeline-stack.ts
import { SecretValue, Stack, StackProps } from "aws-cdk-lib";
import { Construct } from "constructs";
import { Artifact, IStage, Pipeline } from "aws-cdk-lib/aws-codepipeline";
import {
CloudFormationCreateUpdateStackAction,
CodeBuildAction,
CodeBuildActionType,
GitHubSourceAction,
} from "aws-cdk-lib/aws-codepipeline-actions";
import {
BuildEnvironmentVariableType,
BuildSpec,
LinuxBuildImage,
PipelineProject,
} from "aws-cdk-lib/aws-codebuild";
import { SnsTopic } from "aws-cdk-lib/aws-events-targets";
import { Topic } from "aws-cdk-lib/aws-sns";
import { EventField, RuleTargetInput } from "aws-cdk-lib/aws-events";
import { EmailSubscription, SmsSubscription } from "aws-cdk-lib/aws-sns-subscriptions";
import ApiStack from "./stacks/api-stack";
import { ManagedPolicy, Role, ServicePrincipal } from "aws-cdk-lib/aws-iam";
import { SecurityGroup, SubnetType, Vpc } from "aws-cdk-lib/aws-ec2";
import { Secret } from "aws-cdk-lib/aws-ecs";
import { BuildEnvironmentVariable } from "aws-cdk-lib/aws-codebuild/lib/project";
import * as SecretsManager from "aws-cdk-lib/aws-secretsmanager";
import { getApplicationEnvironment, getApplicationSecrets } from "./secrets-helper";
const capFirst = (str: string): string => {
return str.charAt(0).toUpperCase() + str.slice(1);
};
interface PipelineStackProps extends StackProps {
environment: string;
emailAddress: string;
phoneNumber: string;
branch: string;
secrets: {
arn: string;
};
repo: {
uri: string;
name: string;
};
}
export class PipelineStack extends Stack {
private readonly envName: string;
private readonly pipeline: Pipeline;
// source outputs
private cdkSourceOutput: Artifact;
private applicationSourceOutput: Artifact;
// code source actions
private cdkSourceAction: GitHubSourceAction;
private applicationSourceAction: GitHubSourceAction;
// build outputs
private cdkBuildOutput: Artifact;
private applicationBuildOutput: Artifact;
// notifications
private pipelineNotificationsTopic: Topic;
private readonly codeBuildVpc: Vpc;
private readonly codeBuildSecurityGroup: SecurityGroup;
private readonly secrets: SecretsManager.ISecret;
private readonly ecrCodeBuildRole: Role;
// stages
private sourceStage: IStage;
private selfMutateStage: IStage;
private buildStage: IStage;
private apiTestsStage: IStage;
constructor(scope: Construct, id: string, props: PipelineStackProps) {
super(scope, id, props);
this.envName = props.environment;
this.addNotifications(props);
this.ecrCodeBuildRole = new Role(this, "application-build-project-role", {
assumedBy: new ServicePrincipal("codebuild.amazonaws.com"),
managedPolicies: [
ManagedPolicy.fromAwsManagedPolicyName("AmazonEC2ContainerRegistryPowerUser"),
],
});
this.codeBuildVpc = new Vpc(this, "codebuild-vpc", {
vpcName: "codebuild-vpc",
enableDnsSupport: true,
});
this.codeBuildSecurityGroup = new SecurityGroup(this, "codebuild-vpc-security-group", {
vpc: this.codeBuildVpc,
allowAllOutbound: true,
});
this.secrets = SecretsManager.Secret.fromSecretCompleteArn(this, "secrets", props.secrets.arn);
this.pipeline = new Pipeline(this, "pipeline", {
pipelineName: `${capFirst(this.envName)}Pipeline`,
crossAccountKeys: false,
restartExecutionOnUpdate: true,
});
// STAGE 1 - Source Stage
this.addSourceStage(props);
// STAGE 2 - Build Stage
this.addBuildStage(props);
// STAGE 3: SelfMutate Stage
this.addSelfMutateStage();
// STAGE 4: Testing
this.addTestStage();
}
addNotifications(props: PipelineStackProps) {
this.pipelineNotificationsTopic = new Topic(this, "pipeline-notifications-topic", {
topicName: `PipelineNotifications${capFirst(props.environment)}`,
});
this.pipelineNotificationsTopic.addSubscription(new EmailSubscription(props.emailAddress));
this.pipelineNotificationsTopic.addSubscription(new SmsSubscription(props.phoneNumber));
}
/**
* Stage 1
*/
addSourceStage(props: PipelineStackProps) {
this.cdkSourceOutput = new Artifact("cdk-source-output");
this.cdkSourceAction = new GitHubSourceAction({
actionName: "CdkSource",
owner: "my-org",
repo: "my-cdk-repo",
branch: "main",
oauthToken: SecretValue.secretsManager("/connections/github/access-token"),
output: this.cdkSourceOutput,
});
this.applicationSourceOutput = new Artifact("ApplicationSourceOutput");
this.applicationSourceAction = new GitHubSourceAction({
actionName: "ApplicationSource",
owner: "my-org",
repo: "my-application-repo",
branch: props.branch,
oauthToken: SecretValue.secretsManager("/connections/github/access-token"),
output: this.applicationSourceOutput,
});
this.sourceStage = this.pipeline.addStage({
stageName: "Source",
actions: [this.cdkSourceAction, this.applicationSourceAction],
});
}
/**
* stage 2
*/
addBuildStage(props: PipelineStackProps) {
const cdkBuildAction = this.createCdkBuildAction();
const applicationBuildAction = this.createApplicationBuildAction(props);
this.buildStage = this.pipeline.addStage({
stageName: "Build",
actions: [cdkBuildAction, applicationBuildAction],
});
}
/**
* stage 3
*/
addSelfMutateStage() {
this.selfMutateStage = this.pipeline.addStage({
stageName: "PipelineUpdate",
actions: [
new CloudFormationCreateUpdateStackAction({
actionName: "PipelineCreateUpdateStackAction",
stackName: this.stackName,
templatePath: this.cdkBuildOutput.atPath(`${this.stackName}.template.json`),
adminPermissions: true,
}),
],
});
}
/**
* stage 4
*/
addTestStage() {
const testAction = new CodeBuildAction({
actionName: "RunApiTests",
type: CodeBuildActionType.TEST,
input: this.applicationSourceOutput,
project: new PipelineProject(this, "api-tests-project", {
vpc: this.codeBuildVpc,
securityGroups: [this.codeBuildSecurityGroup],
environment: {
buildImage: LinuxBuildImage.STANDARD_5_0,
privileged: true,
},
buildSpec: BuildSpec.fromObject({
version: "0.2",
phases: {
install: {
commands: ["cp .env.testing .env"],
},
build: {
commands: [
"ls",
"docker-compose -f docker-compose.staging.yml run -e NODE_ENV=testing --rm api node ace test",
],
},
},
}),
}),
runOrder: 1,
});
this.apiTestsStage = this.pipeline.addStage({
stageName: "RunApiTests",
actions: [testAction],
});
}
createCdkBuildAction() {
this.cdkBuildOutput = new Artifact("CdkBuildOutput");
return new CodeBuildAction({
actionName: "CdkBuildAction",
input: this.cdkSourceOutput,
outputs: [this.cdkBuildOutput],
project: new PipelineProject(this, "cdk-build-project", {
environment: {
buildImage: LinuxBuildImage.STANDARD_5_0,
},
buildSpec: BuildSpec.fromSourceFilename("build-specs/cdk-build-spec.yml"),
}),
});
}
createApplicationBuildAction(props: PipelineStackProps) {
this.applicationBuildOutput = new Artifact("ApplicationBuildOutput");
const project = new PipelineProject(this, "application-build-project", {
vpc: this.codeBuildVpc,
securityGroups: [this.codeBuildSecurityGroup],
environment: {
buildImage: LinuxBuildImage.STANDARD_5_0,
privileged: true,
},
environmentVariables: {
ENV: {
value: this.envName,
},
ECR_REPO_URI: {
value: props.repo.uri,
},
ECR_REPO_NAME: {
value: props.repo.name,
},
AWS_REGION: {
value: props.env!.region,
},
},
buildSpec: BuildSpec.fromObject({
version: "0.2",
phases: {
pre_build: {
commands: [
"echo 'Logging into Amazon ECR...'",
"aws ecr get-login-password --region $AWS_REGION | docker login --username AWS --password-stdin $ECR_REPO_URI",
'COMMIT_HASH=$(echo "$CODEBUILD_RESOLVED_SOURCE_VERSION" | head -c 8)',
],
},
build: {
commands: ["docker build -t $ECR_REPO_NAME:latest ."],
},
post_build: {
commands: [
"docker tag $ECR_REPO_NAME:latest $ECR_REPO_URI/$ECR_REPO_NAME:latest",
"docker tag $ECR_REPO_NAME:latest $ECR_REPO_URI/$ECR_REPO_NAME:$ENV-$COMMIT_HASH",
"docker push $ECR_REPO_URI/$ECR_REPO_NAME:latest",
"docker push $ECR_REPO_URI/$ECR_REPO_NAME:$ENV-$COMMIT_HASH",
],
},
},
}),
role: this.ecrCodeBuildRole,
});
return new CodeBuildAction({
actionName: "ApplicationBuildAction",
input: this.applicationSourceOutput,
outputs: [this.applicationBuildOutput],
project: project,
});
}
public addDatabaseMigrationStage(apiStack: ApiStack, stageName: string): IStage {
let buildEnv: { [name: string]: BuildEnvironmentVariable } = {
ENV: {
value: this.envName,
},
ECR_REPO_URI: {
type: BuildEnvironmentVariableType.PLAINTEXT,
value: apiStack.repoUri,
},
ECR_REPO_NAME: {
type: BuildEnvironmentVariableType.PLAINTEXT,
value: apiStack.repoName,
},
AWS_REGION: {
type: BuildEnvironmentVariableType.PLAINTEXT,
value: this.region,
},
};
buildEnv = this.getBuildEnvAppSecrets(getApplicationSecrets(this.secrets), buildEnv);
buildEnv = this.getBuildEnvAppEnvVars(
getApplicationEnvironment({
REDIS_HOST: apiStack.redisHost.importValue,
REDIS_PORT: apiStack.redisPort.importValue,
}),
buildEnv,
);
let envVarNames = Object.keys(buildEnv);
const envFileCommand = `printenv | grep '${envVarNames.join("\\|")}' >> .env`;
return this.pipeline.addStage({
stageName: stageName,
actions: [
new CodeBuildAction({
actionName: "DatabaseMigrations",
input: this.applicationSourceOutput,
project: new PipelineProject(this, "database-migrations-project", {
description: "Run database migrations against RDS database",
environment: {
buildImage: LinuxBuildImage.STANDARD_5_0,
privileged: true,
},
environmentVariables: buildEnv,
buildSpec: BuildSpec.fromObject({
version: "0.2",
phases: {
pre_build: {
commands: [
"echo 'Logging into Amazon ECR...'",
"aws ecr get-login-password --region $AWS_REGION | docker login --username AWS --password-stdin $ECR_REPO_URI",
'COMMIT_HASH=$(echo "$CODEBUILD_RESOLVED_SOURCE_VERSION" | head -c 8)',
envFileCommand,
"cat .env",
],
},
build: {
commands: [
`docker run --env-file .env --name api $ECR_REPO_URI/$ECR_REPO_NAME:$ENV-$COMMIT_HASH node ace migration:run --force`,
": > .env",
],
},
},
}),
role: this.ecrCodeBuildRole,
}),
}),
],
});
}
private getBuildEnvAppSecrets(
secrets: { [key: string]: Secret },
buildEnv: { [name: string]: BuildEnvironmentVariable },
): { [name: string]: BuildEnvironmentVariable } {
for (let key in secrets) {
buildEnv[key] = {
type: BuildEnvironmentVariableType.SECRETS_MANAGER,
value: `${this.secrets.secretArn}:${key}`,
};
}
return buildEnv;
}
private getBuildEnvAppEnvVars(
vars: { [key: string]: string },
buildEnv: { [name: string]: BuildEnvironmentVariable },
): { [name: string]: BuildEnvironmentVariable } {
for (let key in vars) {
buildEnv[key] = {
value: vars[key],
};
}
return buildEnv;
}
public addApplicationStage(apiStack: ApiStack, stageName: string): IStage {
return this.pipeline.addStage({
stageName: stageName,
actions: [
new CloudFormationCreateUpdateStackAction({
actionName: "ApplicationUpdate",
stackName: apiStack.stackName,
templatePath: this.cdkBuildOutput.atPath(`${apiStack.stackName}.template.json`),
adminPermissions: true,
}),
],
});
}
}
// api-stack.ts
import { CfnOutput, CfnResource, Lazy, Stack, StackProps } from "aws-cdk-lib";
import * as EC2 from "aws-cdk-lib/aws-ec2";
import { ISubnet } from "aws-cdk-lib/aws-ec2";
import * as ECS from "aws-cdk-lib/aws-ecs";
import { DeploymentControllerType, ScalableTaskCount } from "aws-cdk-lib/aws-ecs";
import * as EcsPatterns from "aws-cdk-lib/aws-ecs-patterns";
import * as RDS from "aws-cdk-lib/aws-rds";
import { Credentials } from "aws-cdk-lib/aws-rds";
import * as Route53 from "aws-cdk-lib/aws-route53";
import * as Route53Targets from "aws-cdk-lib/aws-route53-targets";
import * as ECR from "aws-cdk-lib/aws-ecr";
import * as CertificateManager from "aws-cdk-lib/aws-certificatemanager";
import * as SecretsManager from "aws-cdk-lib/aws-secretsmanager";
import * as ElasticCache from "aws-cdk-lib/aws-elasticache";
import { Construct } from "constructs";
import { getApplicationEnvironment, getApplicationSecrets } from "../secrets-helper";
export type ApiStackProps = StackProps & {
environment: string;
hostedZone: {
id: string;
name: string;
};
domainName: string;
scaling: {
desiredCount: number;
maxCount: number;
cpuPercentage: number;
memoryPercentage: number;
};
repository: {
uri: string;
arn: string;
name: string;
};
secrets: { arn: string };
};
export default class ApiStack extends Stack {
vpc: EC2.Vpc;
cluster: ECS.Cluster;
ecsService: EcsPatterns.ApplicationLoadBalancedFargateService;
certificate: CertificateManager.ICertificate;
repository: ECR.IRepository;
database: RDS.IDatabaseInstance;
databaseCredentials: Credentials;
hostedZone: Route53.IHostedZone;
aliasRecord: Route53.ARecord;
redis: ElasticCache.CfnReplicationGroup;
repoUri: string;
repoName: string;
applicationEnvVariables: {
[key: string]: string;
};
redisHost: CfnOutput;
redisPort: CfnOutput;
gatewayUrl: CfnOutput;
constructor(scope: Construct, id: string, props: ApiStackProps) {
super(scope, id, props);
this.repoUri = props.repository.uri;
this.repoName = props.repository.name;
this.setUpVpc(props);
this.setUpRedisCluster(props);
this.setUpDatabase(props);
this.setUpCluster(props);
this.setUpHostedZone(props);
this.setUpCertificate(props);
this.setUpRepository(props);
this.setUpEcsService(props);
this.setUpAliasRecord(props);
}
private resourceName(props: ApiStackProps, resourceType: string): string {
return `twibs-api-${resourceType}-${props.environment}`;
}
private setUpVpc(props: ApiStackProps) {
this.vpc = new EC2.Vpc(this, this.resourceName(props, "vpc"), {
maxAzs: 3, // Default is all AZs in region
});
}
private setUpRedisCluster(props: ApiStackProps) {
const subnetGroup = new ElasticCache.CfnSubnetGroup(this, "cache-subnet-group", {
cacheSubnetGroupName: "redis-cache-subnet-group",
subnetIds: this.vpc.privateSubnets.map((subnet: ISubnet) => subnet.subnetId),
description: "Subnet group for Redis Cache cluster",
});
const securityGroup = new EC2.SecurityGroup(this, "redis-security-group", {
vpc: this.vpc,
description: `SecurityGroup associated with RedisDB Cluster - ${props.environment}`,
allowAllOutbound: false,
});
securityGroup.addIngressRule(
EC2.Peer.ipv4(this.vpc.vpcCidrBlock),
EC2.Port.tcp(6379),
"Allow from VPC on port 6379",
);
this.redis = new ElasticCache.CfnReplicationGroup(this, "redis", {
numNodeGroups: 1,
cacheNodeType: "cache.t2.small",
engine: "redis",
multiAzEnabled: false,
autoMinorVersionUpgrade: false,
cacheParameterGroupName: "default.redis6.x.cluster.on",
engineVersion: "6.x",
cacheSubnetGroupName: subnetGroup.ref,
securityGroupIds: [securityGroup.securityGroupId],
replicationGroupDescription: "RedisDB setup by CDK",
replicasPerNodeGroup: 0,
port: 6379,
});
}
private setUpDatabase(props: ApiStackProps) {
if (["production", "staging", "develop"].includes(props.environment)) {
return;
}
this.databaseCredentials = Credentials.fromUsername("my_db_username");
this.database = new RDS.DatabaseInstance(this, "database", {
vpc: this.vpc,
engine: RDS.DatabaseInstanceEngine.postgres({
version: RDS.PostgresEngineVersion.VER_13_4,
}),
credentials: this.databaseCredentials,
databaseName: `my_app_${props.environment}`,
deletionProtection: true,
});
}
private setUpCluster(props: ApiStackProps) {
this.cluster = new ECS.Cluster(this, this.resourceName(props, "cluster"), {
vpc: this.vpc,
capacity: {
instanceType: EC2.InstanceType.of(EC2.InstanceClass.T3, EC2.InstanceSize.`SMALL`),
},
});
}
private setUpHostedZone(props: ApiStackProps) {
this.hostedZone = Route53.HostedZone.fromHostedZoneAttributes(
this,
this.resourceName(props, "hosted-zone"),
{
hostedZoneId: props.hostedZone.id,
zoneName: props.hostedZone.name,
},
);
}
private setUpCertificate(props: ApiStackProps) {
this.certificate = new CertificateManager.Certificate(this, "certificate", {
domainName: props.domainName,
validation: CertificateManager.CertificateValidation.fromDns(this.hostedZone),
});
}
private setUpRepository(props: ApiStackProps) {
this.repository = ECR.Repository.fromRepositoryAttributes(
this,
this.resourceName(props, "repository"),
{
repositoryArn: props.repository.arn,
repositoryName: props.repository.name,
},
);
}
private setUpEcsService(props: ApiStackProps) {
const secrets = SecretsManager.Secret.fromSecretCompleteArn(this, "secrets", props.secrets.arn);
this.redisHost = new CfnOutput(this, "redis-host-output", {
value: this.redis.attrConfigurationEndPointAddress,
exportName: "redis-host-output",
});
this.redisPort = new CfnOutput(this, "redis-port-output", {
value: this.redis.attrConfigurationEndPointPort,
exportName: "redis-port-output",
});
// Create a load-balanced ecs-service service and make it public
this.ecsService = new EcsPatterns.ApplicationLoadBalancedFargateService(
this,
this.resourceName(props, "ecs-service"),
{
serviceName: `${props.environment}-api-service`,
cluster: this.cluster, // Required
cpu: 256, // Default is 256
desiredCount: props.scaling.desiredCount, // Default is 1
taskImageOptions: {
image: ECS.ContainerImage.fromEcrRepository(this.repository),
environment: getApplicationEnvironment({
REDIS_HOST: this.redis.attrConfigurationEndPointAddress,
REDIS_PORT: this.redis.attrConfigurationEndPointPort,
}),
secrets: getApplicationSecrets(secrets),
},
memoryLimitMiB: 512, // Default is 512
publicLoadBalancer: true, // Default is false
domainZone: this.hostedZone,
certificate: this.certificate,
},
);
const scalableTarget = this.ecsService.service.autoScaleTaskCount({
minCapacity: props.scaling.desiredCount,
maxCapacity: props.scaling.maxCount,
});
scalableTarget.scaleOnCpuUtilization("cpu-scaling", {
targetUtilizationPercent: props.scaling.cpuPercentage,
});
scalableTarget.scaleOnMemoryUtilization("memory-scaling", {
targetUtilizationPercent: props.scaling.memoryPercentage,
});
secrets.grantRead(this.ecsService.taskDefinition.taskRole);
}
private setUpAliasRecord(props: ApiStackProps) {
this.gatewayUrl = new CfnOutput(this, "gateway-url-output", {
value: this.ecsService.loadBalancer.loadBalancerDnsName,
});
this.aliasRecord = new Route53.ARecord(this, "alias-record", {
zone: this.hostedZone,
recordName: props.domainName,
target: Route53.RecordTarget.fromAlias(
new Route53Targets.LoadBalancerTarget(this.ecsService.loadBalancer),
),
});
const shouldCreateWWW = props.domainName.split(".").length === 2;
if (shouldCreateWWW) {
new Route53.ARecord(this, "alias-record-www", {
zone: this.hostedZone,
recordName: `www.${props.domainName}`,
target: Route53.RecordTarget.fromAlias(
new Route53Targets.LoadBalancerTarget(this.ecsService.loadBalancer),
),
});
}
}
}
Any advice is greatly appreciated.

Answer 1

I would strongly advise moving from the Docker directory to ECR public gallery to avoid ratelimit issues: https://gallery.ecr.aws/
That said, to answer the question about the number of NATs created. As you can see in the CDK docs, what you're seeing reflects the default behavior (emphasis mine):
A VPC consists of one or more subnets that instances can be placed
into. CDK distinguishes three different subnet types:
Public (SubnetType.PUBLIC) - public subnets connect directly to the Internet using an Internet Gateway. If you want your instances to
have a public IP address and be directly reachable from the Internet,
you must place them in a public subnet.
Private with Internet Access (SubnetType.PRIVATE_WITH_NAT) - instances in private subnets are not directly routable from the
Internet, and connect out to the Internet via a NAT gateway. By
default, a NAT gateway is created in every public subnet for maximum
availability. Be aware that you will be charged for NAT gateways.
Isolated (SubnetType.PRIVATE_ISOLATED) - isolated subnets do not route from or to the Internet, and as such do not require NAT
gateways. They can only connect to or be connected to from other
instances in the same VPC. A default VPC configuration will not
include isolated subnets,
A default VPC configuration will create public and private subnets.
However, if natGateways:0 and subnetConfiguration is undefined,
default VPC configuration will create public and isolated subnets.
So a separate NAT is created for every Public subnet.
Also, the docs for the natGateways parameter mentioned above also describe the default behavior:
(default: One NAT gateway/instance per Availability Zone)
To limit the number of AZs used by the VPC, specify the maxAzs parameter. Set it to 1 to only have a single NAT per VPC.
If you're fine with making the resources in the VPC publicly reachable from the internet, you can place them in Public subnets and avoid the creation of NATs altogether.
this.vpc = new EC2.Vpc(this, this.resourceName(props, "vpc"), {
maxAzs: 1,
natGateways: 0;
});
If you do this, you have to tell your resources to use the public subnet instead of the isolated one.
However, CodeBuild projects do not support this.
They require a NAT to connect to the internet if placed into a VPC. See this question for details.
So if you want your build project to be in a VPC, you need to place it into a private subnet. This is done by default, so no additional configuration needed. Just make sure you have at least one NAT gateway.
To sum up, the real solution to the Docker Hub rate limit issue is to switch over to ECR Public gallery.