Fix: Lambda SnapStart for Python functions doesn't supported in some …

…regions. (#618)

* Fix: Lambda SnapStart for Python functions doesn't supported in some regions.
- Provide the option to disable Lambda SnapStart.
  - `enableLambdaSnapStart` in cdk.json

* Disable Lambda SnapStart on super-easy deployment.

README.md

@@ -90,6 +90,7 @@ chmod +x bin.sh
 You can specify the following parameters during deployment to enhance security and customization:
 
 - **--disable-self-register**: Disable self-registration (default: enabled). If this flag is set, you will need to create all users on cognito and it will not allow users to self register their accounts.
+- **--enable-lambda-snapstart**: Enable [Lambda SnapStart](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html) (default: disabled). If this flag is set, improves cold start times for Lambda functions, providing faster response times for better user experience.
 - **--ipv4-ranges**: Comma-separated list of allowed IPv4 ranges. (default: allow all ipv4 addresses)
 - **--ipv6-ranges**: Comma-separated list of allowed IPv6 ranges. (default: allow all ipv6 addresses)
 - **--disable-ipv6**: Disable connections over IPv6. (default: enabled)
@@ -176,6 +177,7 @@ cdk bootstrap aws://<account id>/us-east-1
 
   - `bedrockRegion`: Region where Bedrock is available. **NOTE: Bedrock does NOT support all regions for now.**
   - `allowedIpV4AddressRanges`, `allowedIpV6AddressRanges`: Allowed IP Address range.
+  - `enableLambdaSnapStart`: Defaults to true. Set to false if deploying to a [region that doesn't support Lambda SnapStart for Python functions](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html#snapstart-supported-regions).
 
 - Deploy this sample project
 
@@ -288,6 +290,14 @@ This is an account/region-level setting, affecting the entire application rather
 "enableBedrockCrossRegionInference": true
 ```
 
+### Lambda SnapStart
+
+[Lambda SnapStart](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html) improves cold start times for Lambda functions, providing faster response times for better user experience. On the other hand, for Python functions, there is a [charge depending on cache size](https://aws.amazon.com/lambda/pricing/#SnapStart_Pricing) and [not available in some regions](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html#snapstart-supported-regions) currently. To disable SnapStart, edit `cdk.json`.
+
+```json
+"enableLambdaSnapStart": false
+```
+
 ### Local Development
 
 See [LOCAL DEVELOPMENT](./docs/LOCAL_DEVELOPMENT.md).

bin.sh

@@ -35,6 +35,7 @@ done
 
 # Default parameters
 ALLOW_SELF_REGISTER="true"
+ENABLE_LAMBDA_SNAPSTART="false"
 IPV4_RANGES=""
 IPV6_RANGES=""
 DISABLE_IPV6="false"
@@ -46,6 +47,7 @@ VERSION="v2"
 while [[ "$#" -gt 0 ]]; do
     case $1 in
         --disable-self-register) ALLOW_SELF_REGISTER="false" ;;
+        --enable-lambda-snapstart) ENABLE_LAMBDA_SNAPSTART="true" ;;
         --disable-ipv6) DISABLE_IPV6="true" ;;
         --ipv4-ranges) IPV4_RANGES="$2"; shift ;;
         --ipv6-ranges) IPV6_RANGES="$2"; shift ;;
@@ -74,6 +76,7 @@ aws cloudformation deploy \
   --capabilities CAPABILITY_IAM \
   --parameter-overrides \
     AllowSelfRegister=$ALLOW_SELF_REGISTER \
+    EnableLambdaSnapStart=$ENABLE_LAMBDA_SNAPSTART \
     DisableIpv6=$DISABLE_IPV6 \
     Ipv4Ranges="$IPV4_RANGES" \
     Ipv6Ranges="$IPV6_RANGES" \

cdk/bin/bedrock-chat.ts

@@ -43,6 +43,7 @@ const USE_STAND_BY_REPLICAS: boolean =
 const ENABLE_BEDROCK_CROSS_REGION_INFERENCE: boolean = app.node.tryGetContext(
   "enableBedrockCrossRegionInference"
 );
+const ENABLE_LAMBDA_SNAPSTART: boolean = app.node.tryGetContext("enableLambdaSnapStart");
 
 // WAF for frontend
 // 2023/9: Currently, the WAF for CloudFront needs to be created in the North America region (us-east-1), so the stacks are separated
@@ -94,6 +95,7 @@ const chat = new BedrockChatStack(app, `BedrockChatStack`, {
   documentBucket: bedrockRegionResources.documentBucket,
   useStandbyReplicas: USE_STAND_BY_REPLICAS,
   enableBedrockCrossRegionInference: ENABLE_BEDROCK_CROSS_REGION_INFERENCE,
+  enableLambdaSnapStart: ENABLE_LAMBDA_SNAPSTART,
 });
 chat.addDependency(waf);
 chat.addDependency(bedrockRegionResources);

cdk/cdk.json

@@ -69,6 +69,7 @@
       "8000:0000:0000:0000:0000:0000:0000:0000/1"
     ],
     "enableRagReplicas": true,
-    "enableBedrockCrossRegionInference": true
+    "enableBedrockCrossRegionInference": true,
+    "enableLambdaSnapStart": true
   }
 }

cdk/lib/bedrock-chat-stack.ts

@@ -39,6 +39,7 @@ export interface BedrockChatStackProps extends StackProps {
   readonly documentBucket: Bucket;
   readonly useStandbyReplicas: boolean;
   readonly enableBedrockCrossRegionInference: boolean;
+  readonly enableLambdaSnapStart: boolean;
 }
 
 export class BedrockChatStack extends cdk.Stack {
@@ -165,6 +166,7 @@ export class BedrockChatStack extends cdk.Stack {
       usageAnalysis,
       largeMessageBucket,
       enableMistral: props.enableMistral,
+      enableLambdaSnapStart: props.enableLambdaSnapStart,
     });
     props.documentBucket.grantReadWrite(backendApi.handler);
 
@@ -181,6 +183,7 @@ export class BedrockChatStack extends cdk.Stack {
       enableMistral: props.enableMistral,
       enableBedrockCrossRegionInference:
         props.enableBedrockCrossRegionInference,
+      enableLambdaSnapStart: props.enableLambdaSnapStart,
     });
     frontend.buildViteApp({
       backendApiEndpoint: backendApi.api.apiEndpoint,

cdk/lib/constructs/api.ts

@@ -4,9 +4,7 @@ import { ITable } from "aws-cdk-lib/aws-dynamodb";
 import { HttpLambdaIntegration } from "aws-cdk-lib/aws-apigatewayv2-integrations";
 import { HttpUserPoolAuthorizer } from "aws-cdk-lib/aws-apigatewayv2-authorizers";
 import {
-  Code,
-  DockerImageCode,
-  DockerImageFunction,
+  Architecture,
   IFunction,
   LayerVersion,
   Runtime,
@@ -18,7 +16,6 @@ import {
   HttpMethod,
 } from "aws-cdk-lib/aws-apigatewayv2";
 import { Auth } from "./auth";
-import { Platform } from "aws-cdk-lib/aws-ecr-assets";
 import { Stack } from "aws-cdk-lib";
 import * as iam from "aws-cdk-lib/aws-iam";
 import * as logs from "aws-cdk-lib/aws-logs";
@@ -41,6 +38,7 @@ export interface ApiProps {
   readonly bedrockCustomBotProject: codebuild.IProject;
   readonly usageAnalysis?: UsageAnalysis;
   readonly enableMistral: boolean;
+  readonly enableLambdaSnapStart: boolean;
 }
 
 export class Api extends Construct {
@@ -184,6 +182,7 @@ export class Api extends Construct {
         buildArgs: { POETRY_VERSION: "1.8.3" },
       },
       runtime: Runtime.PYTHON_3_12,
+      architecture: Architecture.X86_64,
       memorySize: 1024,
       timeout: Duration.minutes(15),
       environment: {
@@ -212,7 +211,7 @@ export class Api extends Construct {
       },
       role: handlerRole,
       logRetention: logs.RetentionDays.THREE_MONTHS,
-      snapStart: SnapStartConf.ON_PUBLISHED_VERSIONS,
+      snapStart: props.enableLambdaSnapStart ? SnapStartConf.ON_PUBLISHED_VERSIONS : undefined,
       layers: [
         LayerVersion.fromLayerVersionArn(
           this,

cdk/lib/constructs/websocket.ts

@@ -29,6 +29,7 @@ export interface WebSocketProps {
   readonly accessLogBucket?: s3.Bucket;
   readonly enableMistral: boolean;
   readonly enableBedrockCrossRegionInference: boolean;
+  readonly enableLambdaSnapStart: boolean;
 }
 
 export class WebSocket extends Construct {
@@ -110,7 +111,7 @@ export class WebSocket extends Construct {
           props.enableBedrockCrossRegionInference.toString(),
       },
       role: handlerRole,
-      snapStart: SnapStartConf.ON_PUBLISHED_VERSIONS,
+      snapStart: props.enableLambdaSnapStart ? SnapStartConf.ON_PUBLISHED_VERSIONS : undefined,
       logRetention: logs.RetentionDays.THREE_MONTHS,
     });
 

cdk/test/cdk.test.ts

@@ -55,6 +55,7 @@ describe("Bedrock Chat Stack Test", () => {
         documentBucket: bedrockRegionResourcesStack.documentBucket,
         useStandbyReplicas: false,
         enableBedrockCrossRegionInference: false,
+        enableLambdaSnapStart: true,
       }
     );
     const hasGoogleProviderTemplate = Template.fromStack(
@@ -125,6 +126,7 @@ describe("Bedrock Chat Stack Test", () => {
         documentBucket: bedrockRegionResourcesStack.documentBucket,
         useStandbyReplicas: false,
         enableBedrockCrossRegionInference: false,
+        enableLambdaSnapStart: true,
       }
     );
     const hasOidcProviderTemplate = Template.fromStack(hasOidcProviderStack);
@@ -185,6 +187,7 @@ describe("Bedrock Chat Stack Test", () => {
       documentBucket: bedrockRegionResourcesStack.documentBucket,
       useStandbyReplicas: false,
       enableBedrockCrossRegionInference: false,
+      enableLambdaSnapStart: true,
     });
     const template = Template.fromStack(stack);
 

deploy.yml

@@ -5,6 +5,9 @@ Parameters:
   AllowSelfRegister:
     Type: String
     Default: "true"
+  EnableLambdaSnapStart:
+    Type: String
+    Default: "false"
   Ipv4Ranges:
     Type: String
     Default: '["0.0.0.0/1","128.0.0.0/1"]' # Set default values based on current config
@@ -113,6 +116,8 @@ Resources:
         EnvironmentVariables:
           - Name: ALLOW_SELF_REGISTER
             Value: !Ref AllowSelfRegister
+          - Name: ENABLE_LAMBDA_SNAPSTART
+            Value: !Ref EnableLambdaSnapStart
           - Name: IPV4_RANGES
             Value: !Ref Ipv4Ranges
           - Name: IPV6_RANGES
@@ -148,7 +153,8 @@ Resources:
                   "echo 'Build phase...'",
                   "git clone --branch $VERSION https://github.com/aws-samples/bedrock-claude-chat.git",
                   "cd bedrock-claude-chat",
-                  "if [ \"$ALLOW_SELF_REGISTER\" = \"false\" ]; then sed -i 's/\"selfSignUpEnabled\": true,/\"selfSignUpEnabled\": false,/' cdk/cdk.json; fi",
+                  "if [ \"$ALLOW_SELF_REGISTER\" = \"false\" ]; then sed -i 's/\"selfSignUpEnabled\": true/\"selfSignUpEnabled\": false/' cdk/cdk.json; fi",
+                  "if [ \"$ENABLE_LAMBDA_SNAPSTART\" = \"false\" ]; then sed -i 's/\"enableLambdaSnapStart\": true/\"enableLambdaSnapStart\": false/' cdk/cdk.json; fi",
                   "if [ ! -z \"$IPV4_RANGES\" ]; then jq --arg ipv4 \"$IPV4_RANGES\" '.context.allowedIpV4AddressRanges = ($ipv4 | split(\",\"))' cdk/cdk.json > temp.json && mv temp.json cdk/cdk.json; fi",
                   "if [ \"$DISABLE_IPV6\" = \"true\" ]; then jq '.context.allowedIpV6AddressRanges = []' cdk/cdk.json > temp.json && mv temp.json cdk/cdk.json; elif [ ! -z \"$IPV6_RANGES\" ]; then jq --arg ipv6 \"$IPV6_RANGES\" '.context.allowedIpV6AddressRanges = ($ipv6 | split(\",\"))' cdk/cdk.json > temp.json && mv temp.json cdk/cdk.json; fi",
                   "if [ ! -z \"$ALLOWED_SIGN_UP_EMAIL_DOMAINS\" ]; then jq --arg domains \"$ALLOWED_SIGN_UP_EMAIL_DOMAINS\" '.context.allowedSignUpEmailDomains = ($domains | split(\",\"))' cdk/cdk.json > temp.json && mv temp.json cdk/cdk.json; fi",

docs/README_ja.md

@@ -60,6 +60,7 @@ chmod +x bin.sh
 デプロイ時に以下のパラメータを指定することで、セキュリティとカスタマイズを強化できます。
 
 - **--disable-self-register**: セルフ登録を無効にします（デフォルト: 有効）。このフラグを設定すると、Cognito 上で全てのユーザーを作成する必要があり、ユーザーが自分でアカウントを登録することはできなくなります。
+- **--enable-lambda-snapstart**: [Lambda SnapStart](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html) を有効化します (デフォルト: 無効)。 このフラグを設定すると、Lambda 関数のコールドスタート時間を短縮し、レスポンスタイムの改善によってユーザー体験を向上させます。
 - **--ipv4-ranges**: 許可する IPv4 範囲のカンマ区切りリスト。（デフォルト: 全ての IPv4 アドレスを許可）
 - **--ipv6-ranges**: 許可する IPv6 範囲のカンマ区切りリスト。（デフォルト: 全ての IPv6 アドレスを許可）
 - **--disable-ipv6**: IPv6 での接続を無効にします (デフォルト: 有効)
@@ -145,6 +146,7 @@ cdk bootstrap aws://<account id>/ap-northeast-1
 
   - `bedrockRegion`: Bedrock が利用できるリージョン
   - `allowedIpV4AddressRanges`, `allowedIpV6AddressRanges`: 許可する IP アドレス範囲の指定
+  - `enableLambdaSnapStart`: デフォルトでは true ですが、[Python 関数の Lambda SnapStart をサポートしていないリージョン](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html#snapstart-supported-regions)にデプロイする場合は false に変更してください。
 
 - プロジェクトをデプロイします
 
@@ -257,6 +259,14 @@ cli および CDK を利用されている場合、`cdk destroy`を実行して
 "enableBedrockCrossRegionInference": true
 ```
 
+### Lambda SnapStart
+
+[Lambda SnapStart](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html) は Lambda 関数のコールドスタート時間を短縮し、レスポンスタイムの改善によってユーザー体験を向上させます。ただし、Python 関数については[キャッシュサイズに比例した利用料金](https://aws.amazon.com/lambda/pricing/#SnapStart_Pricing)が発生するのに加え、[一部のリージョン](https://docs.aws.amazon.com/lambda/latest/dg/snapstart.html#snapstart-supported-regions)では現在利用できません。SnapStart を無効化するには、`cdk.json` を以下のように編集します。
+
+```json
+"enableLambdaSnapStart": false
+```
+
 ### ローカルでの開発について
 
 - [こちら](./LOCAL_DEVELOPMENT_ja.md)を参照ください。