Windows26: [New post] Deployment strategies

Respond to this post by replying above this line

Deployment strategies

by Osama Mustafa

All-at-once

All-at-once deployments instantly shift traffic from the original (old) Lambda function to the updated (new) Lambda function, all at one time. All-at-once deployments can be beneficial when the speed of your deployments matters. In this strategy, the new version of your code is released quickly, and all your users get to access it immediately.

Canary

A linear deployment is similar to canary deployment. In this strategy, you direct a small amount of traffic to

In a canary deployment, you deploy your new version of your application code and shift a small percentage of production traffic to point to that new version. After you have validated that this version is safe and not causing errors, you direct all traffic to the new version of your code.

Linear

A linear deployment is similar to canary deployment. In this strategy, you direct a small amount of traffic to your new version of code at first. After a specified period of time, you automatically increment the amount of traffic that you send to the new version until you're sending 100% of production traffic.

Comparing deployment strategies

To help you decide which deployment strategy to use for your application, you'll need to consider each option's consumer impact, rollback, event model factors, and deployment speed. The comparison table below illustrates these points.

Deployment	Consumer Impact	Rollback	Event Model Factors	Deployment Speed
All-at-once	All at once	Redeploy older version	Any event model at low concurrency rate	Immediate
Canary/ Linear	1-10% typical initial traffic shift, then phased	Revert 100% of traffic to previous deployment	Better for high-concurrency workloads	Minutes to hours

Deployment preferences with AWS SAM

Traffic shifting with aliases is directly integrated into AWS SAM. If you'd like to use all-at-once, canary, or linear deployments with your Lambda functions, you can embed that directly into your AWS SAM templates. You can do this in the deployment preferences section of the template. AWS CodeDeploy uses the deployment preferences section to manage the function rollout as part of the AWS CloudFormation stack update. SAM has several pre-built deployment preferences you can use to deploy your code. See the table below for examples.

Deployment Preferences Type	Description
Canary10Percent30Minutes	Shifts 10 percent of traffic in the first increment. The remaining 90 percent is deployed 30 minutes later.
Canary10Percent5Minutes	Shifts 10 percent of traffic in the first increment. The remaining 90 percent is deployed 5 minutes later.
Canary10Percent10Minutes	Shifts 10 percent of traffic in the first increment. The remaining 90 percent is deployed 10 minutes later.
Canary10Percent15Minutes	Shifts 10 percent of traffic in the first increment. The remaining 90 percent is deployed 15 minutes later.
Linear10PercentEvery10Minutes	Shifts 10 percent of traffic every 10 minutes until all traffic is shifted.
Linear10PercentEvery1Minute	Shifts 10 percent of traffic every minute until all traffic is shifted.
Linear10PercentEvery2Minutes	Shifts 10 percent of traffic every 2 minutes until all traffic is shifted.
Linear10PercentEvery3Minutes	Shifts 10 percent of traffic every 3 minutes until all traffic is shifted.
AllAtOnce	Shifts all traffic to the updated Lambda functions at once.

Creating a deployment pipeline

When you check a piece of code into source control, you don't want to wait for a human to manually approve it or have each piece of code run through different quality checks. Using a CI/CD pipeline can help automate the steps required to release your software deployment and standardize on a core set of quality checks.