[Backport 1.x] Add ec2 instance options for data and ml nodes

README.md

@@ -51,6 +51,8 @@ In order to deploy both the stacks the user needs to provide a set of required a
 | clientNodeCount (Optional)        | integer | Number of dedicated client nodes, default is 0                                                                                                                                                                                                                                                   |
 | ingestNodeCount (Optional)        | integer | Number of dedicated ingest nodes, default is 0                                                                                                                                                                                                                                                   |
 | mlNodeCount (Optional)            | integer | Number of dedicated machine learning nodes, default is 0                                                                                                                                                                                                                                         |
+| dataInstanceType (Optional)       | string  | EC2 instance type for data node, defaults to r5.large. See options in `lib/opensearch-config/node-config.ts` for available options. E.g., `-c dataInstanceType=m5.xlarge`                                                                                                                        |
+| mlInstanceType (Optional)         | string  | EC2 instance type for ml node, defaults to r5.large. See options in `lib/opensearch-config/node-config.ts` for available options. E.g., `-c mlInstanceType=m5.xlarge`                                                                                                                            |
 | jvmSysProps (Optional)            | string  | A comma-separated list of key=value pairs that will be added to `jvm.options` as JVM system properties.                                                                                                                                                                                          |
 | additionalConfig (Optional)       | string  | Additional opensearch.yml config parameters passed as JSON. e.g., `--context additionalConfig='{"plugins.security.nodes_dn": ["CN=*.example.com, OU=SSL, O=Test, L=Test, C=DE", "CN=node.other.com, OU=SSL, O=Test, L=Test, C=DE"], "plugins.security.nodes_dn_dynamic_config_enabled": false}'` |
 | suffix (Optional)                 | string  | An optional string identifier to be concatenated with infra stack name.                                                                                                                                                                                                                          |

lib/infra/infra-stack.ts

@@ -55,6 +55,8 @@
     readonly mlNodeStorage: number,
     readonly jvmSysPropsString?: string,
     readonly additionalConfig?: string,
+    readonly dataEc2InstanceType: InstanceType,
+    readonly mlEc2InstanceType: InstanceType,
     readonly use50PercentHeap: boolean,
     readonly isInternal: boolean,
 }
@@ -84,7 +86,10 @@
       assumedBy: new ServicePrincipal('ec2.amazonaws.com'),
     });
 
-    const ec2InstanceType = (props.cpuType === AmazonLinuxCpuType.X86_64)
+    const singleNodeInstanceType = (props.cpuType === AmazonLinuxCpuType.X86_64)
+      ? InstanceType.of(InstanceClass.R5, InstanceSize.XLARGE) : InstanceType.of(InstanceClass.R6G, InstanceSize.XLARGE);
+
+    const defaultInstanceType = (props.cpuType === AmazonLinuxCpuType.X86_64)
       ? InstanceType.of(InstanceClass.C5, InstanceSize.XLARGE) : InstanceType.of(InstanceClass.C6G, InstanceSize.XLARGE);
 
     const nlb = new NetworkLoadBalancer(this, 'clusterNlb', {
@@ -113,10 +118,10 @@
     }
 
     if (props.singleNodeCluster) {
       console.log('Single node value is true, creating single node configurations');
       singleNodeInstance = new Instance(this, 'single-node-instance', {
         vpc: props.vpc,
-        instanceType: ec2InstanceType,
+        instanceType: singleNodeInstanceType,
         machineImage: MachineImage.latestAmazonLinux({
           generation: AmazonLinuxGeneration.AMAZON_LINUX_2,
           cpuType: props.cpuType,
@@ -164,7 +169,7 @@
       if (managerAsgCapacity > 0) {
         const managerNodeAsg = new AutoScalingGroup(this, 'managerNodeAsg', {
           vpc: props.vpc,
-          instanceType: ec2InstanceType,
+          instanceType: defaultInstanceType,
           machineImage: MachineImage.latestAmazonLinux({
             generation: AmazonLinuxGeneration.AMAZON_LINUX_2,
             cpuType: props.cpuType,
@@ -196,7 +201,7 @@
 
       const seedNodeAsg = new AutoScalingGroup(this, 'seedNodeAsg', {
         vpc: props.vpc,
-        instanceType: ec2InstanceType,
+        instanceType: (seedConfig === 'seed-manager') ? defaultInstanceType : props.dataEc2InstanceType,
         machineImage: MachineImage.latestAmazonLinux({
           generation: AmazonLinuxGeneration.AMAZON_LINUX_2,
           cpuType: props.cpuType,
@@ -211,7 +216,8 @@
         securityGroup: props.securityGroup,
         blockDevices: [{
           deviceName: '/dev/xvda',
-          volume: BlockDeviceVolume.ebs(50, { deleteOnTermination: true }),
+          // eslint-disable-next-line max-len
+          volume: (seedConfig === 'seed-manager') ? BlockDeviceVolume.ebs(50, { deleteOnTermination: true }) : BlockDeviceVolume.ebs(props.dataNodeStorage, { deleteOnTermination: true }),
         }],
         init: CloudFormationInit.fromElements(...InfraStack.getCfnInitElement(this, clusterLogGroup, props, seedConfig)),
         initOptions: {
@@ -223,7 +229,7 @@
 
       const dataNodeAsg = new AutoScalingGroup(this, 'dataNodeAsg', {
         vpc: props.vpc,
-        instanceType: ec2InstanceType,
+        instanceType: props.dataEc2InstanceType,
         machineImage: MachineImage.latestAmazonLinux({
           generation: AmazonLinuxGeneration.AMAZON_LINUX_2,
           cpuType: props.cpuType,
@@ -253,7 +259,7 @@
       } else {
         clientNodeAsg = new AutoScalingGroup(this, 'clientNodeAsg', {
           vpc: props.vpc,
-          instanceType: ec2InstanceType,
+          instanceType: defaultInstanceType,
           machineImage: MachineImage.latestAmazonLinux({
             generation: AmazonLinuxGeneration.AMAZON_LINUX_2,
             cpuType: props.cpuType,
@@ -284,7 +290,7 @@
       if (props.mlNodeCount > 0) {
         const mlNodeAsg = new AutoScalingGroup(this, 'mlNodeAsg', {
           vpc: props.vpc,
-          instanceType: ec2InstanceType,
+          instanceType: props.mlEc2InstanceType,
           machineImage: MachineImage.latestAmazonLinux({
             generation: AmazonLinuxGeneration.AMAZON_LINUX_2,
             cpuType: props.cpuType,
@@ -409,7 +415,7 @@
 
       fileContent['cluster.name'] = `${scope.stackName}-${scope.account}-${scope.region}`;
 
       console.log(dump(fileContent).toString());
       opensearchConfig = dump(fileContent).toString();
       cfnInitConfig.push(InitCommand.shellCommand(`set -ex;cd opensearch; echo "${opensearchConfig}" > config/opensearch.yml`,
         {

lib/opensearch-config/node-config.ts

@@ -5,6 +5,8 @@
 this file be licensed under the Apache-2.0 license or a
 compatible open source license. */
 
+import { InstanceClass, InstanceSize, InstanceType } from 'aws-cdk-lib/aws-ec2';
+
 export const nodeConfig = new Map<string, object>();
 
 nodeConfig.set('manager', {
@@ -46,3 +48,77 @@
   'node.name': 'ml-node',
   'node.roles': ['ml'],
 });
+
+export enum x64Ec2InstanceType {
+  M5_XLARGE = 'm5.xlarge',
+  M5_2XLARGE = 'm5.2xlarge',
+  C5_LARGE = 'c5.large',
+  C5_XLARGE = 'c5.xlarge',
+  R5_LARGE = 'r5.large',
+  R5_XLARGE = 'r5.xlarge',
+  G5_LARGE = 'g5.large',
+  G5_XLARGE = 'g5.xlarge',
+  INF1_XLARGE = 'inf1.xlarge',
+  INF1_2XLARGE = 'inf1.2xlarge'
+}
+
+export enum arm64Ec2InstanceType {
+  M6G_XLARGE = 'm6g.xlarge',
+  M6G_2XLARGE = 'm6g.2xlarge',
+  C6G_LARGE = 'c6g.large',
+  C6G_XLARGE = 'c6g.xlarge',
+  R6G_LARGE = 'r6g.large',
+  R6G_XLARGE = 'r6g.xlarge',
+  G5G_LARGE = 'g5g.large',
+  G5G_XLARGE = 'g5g.xlarge'
+}
+
+export const getX64InstanceTypes = (instanceType: string) => {
+  switch (instanceType) {
+  case x64Ec2InstanceType.M5_XLARGE:
+    return InstanceType.of(InstanceClass.M5, InstanceSize.XLARGE);
+  case x64Ec2InstanceType.M5_2XLARGE:
+    return InstanceType.of(InstanceClass.M5, InstanceSize.XLARGE2);
+  case x64Ec2InstanceType.C5_LARGE:
+    return InstanceType.of(InstanceClass.C5, InstanceSize.LARGE);
+  case x64Ec2InstanceType.C5_XLARGE:
+    return InstanceType.of(InstanceClass.C5, InstanceSize.XLARGE);
+  case x64Ec2InstanceType.R5_LARGE:
+    return InstanceType.of(InstanceClass.R5, InstanceSize.LARGE);
+  case x64Ec2InstanceType.R5_XLARGE:
+    return InstanceType.of(InstanceClass.R5, InstanceSize.XLARGE);
+  case x64Ec2InstanceType.G5_LARGE:
+    return InstanceType.of(InstanceClass.G5, InstanceSize.LARGE);
+  case x64Ec2InstanceType.G5_XLARGE:
+    return InstanceType.of(InstanceClass.G5, InstanceSize.XLARGE);
+  case x64Ec2InstanceType.INF1_XLARGE:
+    return InstanceType.of(InstanceClass.INF1, InstanceSize.XLARGE);
+  case x64Ec2InstanceType.INF1_2XLARGE:
+    return InstanceType.of(InstanceClass.INF1, InstanceSize.XLARGE2);
+  default:
+    throw new Error(`Invalid instance type provided, please provide any one the following: ${Object.values(x64Ec2InstanceType)}`);
+  }
+};
+
+export const getArm64InstanceTypes = (instanceType: string) => {
+  switch (instanceType) {
+  case arm64Ec2InstanceType.M6G_XLARGE:
+    return InstanceType.of(InstanceClass.M6G, InstanceSize.XLARGE);
+  case arm64Ec2InstanceType.M6G_2XLARGE:
+    return InstanceType.of(InstanceClass.M6G, InstanceSize.XLARGE2);
+  case arm64Ec2InstanceType.C6G_LARGE:
+    return InstanceType.of(InstanceClass.C6G, InstanceSize.LARGE);
+  case arm64Ec2InstanceType.C6G_XLARGE:
+    return InstanceType.of(InstanceClass.C6G, InstanceSize.XLARGE);
+  case arm64Ec2InstanceType.R6G_LARGE:
+    return InstanceType.of(InstanceClass.R6G, InstanceSize.LARGE);
+  case arm64Ec2InstanceType.R6G_XLARGE:
+    return InstanceType.of(InstanceClass.R6G, InstanceSize.XLARGE);
+  case arm64Ec2InstanceType.G5G_LARGE:
+    return InstanceType.of(InstanceClass.G5G, InstanceSize.LARGE);
+  case arm64Ec2InstanceType.G5G_XLARGE:
+    return InstanceType.of(InstanceClass.G5G, InstanceSize.XLARGE);
+  default:
+    throw new Error(`Invalid instance type provided, please provide any one the following: ${Object.values(arm64Ec2InstanceType)}`);
+  }
+};

lib/os-cluster-entrypoint.ts

@@ -8,17 +8,33 @@
 import { Construct } from 'constructs';
 import { Stack, StackProps } from 'aws-cdk-lib';
 import {
-  AmazonLinuxCpuType, IVpc, SecurityGroup, Vpc,
+  AmazonLinuxCpuType, InstanceType, IVpc, SecurityGroup, Vpc,
 } from 'aws-cdk-lib/aws-ec2';
 import { dump } from 'js-yaml';
 import { NetworkStack } from './networking/vpc-stack';
 import { InfraStack } from './infra/infra-stack';
+import {
+  x64Ec2InstanceType, arm64Ec2InstanceType, getX64InstanceTypes, getArm64InstanceTypes,
+} from './opensearch-config/node-config';
 
 enum cpuArchEnum{
     X64='x64',
     ARM64='arm64'
 }
 
+const getInstanceType = (instanceType: string, arch: string) => {
+  if (arch === 'x64') {
+    if (instanceType !== 'undefined') {
+      return getX64InstanceTypes(instanceType);
+    }
+    return getX64InstanceTypes('r5.large');
+  }
+  if (instanceType !== 'undefined') {
+    return getArm64InstanceTypes(instanceType);
+  }
+  return getArm64InstanceTypes('r6g.large');
+};
+
 export class OsClusterEntrypoint {
     public stacks: Stack[] = [];
 
@@ -37,7 +53,11 @@
       let dataNodeStorage: number;
       let mlNodeStorage: number;
       let ymlConfig: string = 'undefined';
+      let dataEc2InstanceType: InstanceType;
+      let mlEc2InstanceType: InstanceType;
 
+      const x64InstanceTypes: string[] = Object.keys(x64Ec2InstanceType);
+      const arm64InstanceTypes: string[] = Object.keys(arm64Ec2InstanceType);
       const vpcId: string = scope.node.tryGetContext('vpcId');
       const securityGroupId = scope.node.tryGetContext('securityGroupId');
       const cidrRange = scope.node.tryGetContext('cidr');
@@ -69,14 +89,22 @@
       const dashboardUrl = `${scope.node.tryGetContext('dashboardsUrl')}`;
 
       const cpuArch = `${scope.node.tryGetContext('cpuArch')}`;
+
+      const dataInstanceType = `${scope.node.tryGetContext('dataInstanceType')}`;
+      const mlInstanceType = `${scope.node.tryGetContext('mlInstanceType')}`;
+
       if (cpuArch.toString() === 'undefined') {
         throw new Error('cpuArch parameter is required. The provided value should be either x64 or arm64, any other value is invalid');
         // @ts-ignore
       } else if (Object.values(cpuArchEnum).includes(cpuArch.toString())) {
         if (cpuArch.toString() === cpuArchEnum.X64) {
           instanceCpuType = AmazonLinuxCpuType.X86_64;
+          dataEc2InstanceType = getInstanceType(dataInstanceType, cpuArch.toString());
+          mlEc2InstanceType = getInstanceType(mlInstanceType, cpuArch.toString());
         } else {
           instanceCpuType = AmazonLinuxCpuType.ARM_64;
+          dataEc2InstanceType = getInstanceType(dataInstanceType, cpuArch.toString());
+          mlEc2InstanceType = getInstanceType(mlInstanceType, cpuArch.toString());
         }
       } else {
         throw new Error('Please provide a valid cpu architecture. The valid value can be either x64 or arm64');
@@ -184,6 +212,8 @@
         clientNodeCount: clientCount,
         cpuArch,
         cpuType: instanceCpuType,
+        dataEc2InstanceType,
+        mlEc2InstanceType,
         dashboardsUrl: dashboardUrl,
         dataNodeCount: dataCount,
         distributionUrl,