Azure Lambda Comparison: Key Differences Between AWS Lambda and Azure Functions

AWS Lambda and Azure Functions are often the introductions to Serverless Computing for several Developers. Before going into the comparison of Azure Lambda, let’s understand what is meant by ‘Serverless Computing’. Earlier, for any kind of computation that you wanted to perform on the Cloud, you needed to provision and maintain servers. However, people realized that this is overkill if the server is required only for performing very small calculations at fixed intervals (think of hourly/daily/weekly crons) or responding to infrequent API requests. Even if the server is utilized for just one minute in the entire day, you’d still pay for all the 24 hours. Serverless computing changed this.

Basically, the Cloud service provider like AWS or Azure said, “Look, we’ll provision and maintain the server. You execute your function (it can be cron triggered or API triggered), and we will charge you only for the time your function executes.” Thus, Serverless Computing is essentially Function-as-a-Service (FaaS). You just need to define your function and define when to execute it. The Cloud service provider takes care of the rest. With it so far? Then let’s understand this better with the comparison of Azure Lambda.

Table of Components

• Prerequisites for Understanding Azure Lambda Comparison
• AWS Lambda vs. Azure Functions
  • Configurability
  • Pricing
  • Triggers
  • Languages
  • Cold Start
  • Deployment
  • Concurrency
  • Exclusivity
  • Code Reusability and Isolation
• Conclusion

Prerequisites for Understanding Azure Lambda Comparison

It will help if you have some past exposure to either AWS Lambda or Azure Functions. Even if you don’t have a practical experience with either of the two, you should be able to understand most of this article on Azure Lambda comparison, although the understanding will be limited because of lack of context.

AWS Lambda vs. Azure Functions

This article compares Azure Lambda based on 9 parameters. This list is not exhaustive, but it will provide you with sufficient insights to evaluate the two for your applications. Let’s get started with the Azure Lambda comparison.

Note: Azure Functions have 3 plans: Consumption, Premium, and App Service. Of these, only the Consumption plan is truly Serverless (pay only for what you use). Others have perpetually warm instances and are charged per hour, irrespective of whether they are being used or not. Therefore, in the comparison that follows, we will only consider the Consumption Plan of Azure Functions.

Here are the 9 important parameters for Azure Lambda comparison.

• Configurability
• Pricing
• Triggers
• Languages
• Cold Start
• Deployment
• Concurrency
• Exclusivity
• Code Reusability and Isolation

Configurability

AWS allows you a lot of flexibility in terms of configuration. You can define functions utilizing memory starting from 128 MB up to 10 GB. The CPU power increases proportionally. At 10 GBs, you get 6 vCPUs, and you can use threading and multiprocessing to make full use of the resources. You can configure a function timeout of a maximum of 900 seconds (15 minutes). While several users want this limit to be increased, there are those who say that increasing this limit further beats the purpose of a Lambda. After all, you are supposed to use Lambda functions for tasks where provisioning a server won’t make much sense.

When it comes to Azure Functions, fortunately, or unfortunately, the configurability is lesser. You can’t define the memory for a function. It is calculated at execution (this can be beneficial to you if you want to leave the memory allocation to the Cloud provider). The maximum memory available to a function is 1.5 GB. Also, the max timeout is 600 seconds (10 minutes).

Pricing

Next up, let’s take a look at the pricing of Azure Lambda. Both AWS Lambda and Azure Functions calculate the pricing under 2 heads: invocation and execution. At the time of writing this article, both of them are charging $0.2 per million invocations (it can vary slightly according to the region). For execution, $0.0000167 per GB per second, whereas Azure Functions charge $0.000016 per GB per second. Recently, AWS introduced ARM-based processors (AWS claims it provides 34% better price-performance), and if you opt for them, you are charged $0.000013 per GB per second. 

Note: In the case of AWS, the memory consumption considered for pricing is the memory defined for the Lambda function. Thus, if you defined 1024 MB (1GB), the cost will be 1 GB*(execution seconds) * (cost per GB per second). Even if your function uses only 300 MB, you will pay for 1 GB. You should therefore optimize the memory allocated for the functions.

In the case of Azure, the memory consumption is calculated dynamically by sampling at regular intervals and rounding off to the nearest 128 MB bucket. Thus, the pricing in the case of Azure better reflects the actual memory consumed by your function.

Triggers

Both AWS Lambda and Azure Functions can be triggered by a host of services on the respective Cloud platforms, apart from HTTP and Timer triggers. For example, Lambda can be triggered by the addition of an object to S3 storage, or an object in an SQS queue can trigger an AWS Lambda and be passed on as the event. It also supports several third-party triggers like Auth0, CleverTap, and Datadog. To get a list of all triggers, you just need to click on ‘Add Trigger’ on the Lambda console. 

Similarly, Azure Functions can be triggered by several Azure services like CosmosDB or blob storage. The complete list of triggers can be found here.

Languages

AWS Lambda supports the following languages: Node.js, Python, Ruby, Java, Go, and NET. You can also define a custom runtime. Azure Functions, on the other hand, support C#, F#, Powershell, Node.js, Java, Python (only on Linux instances, not Windows), and Typescript (supported by transpiling to Javascript). Over here as well, you can define a custom handler if you want to work with Go, Rust, etc. Thus, there’s considerable overlap here, and the popular server-side languages are supported by both Lambda and Azure Functions.

Cold Start

This is one of the biggest differentiators between the two. The pure Serverless plan of Azure Functions (Consumption plan) has a huge cold start problem. The cold start times may even go to tens of seconds. An Azure Function requires a cold start after 20 minutes of inactivity.

On AWS Lambda, the cold start time hardly crosses 1-2 seconds. The high cold start time can make Azure Functions (in the Consumption plan) very unsuitable for acting as APIs and high-frequency crons. You may switch to the Premium or the App Service plans to avoid the cold start in Azure Functions, but, again, those aren’t really Serverless plans. On this particular metric of Azure Lambda comparison, AWS Lambda wins hands down.

Deployment

You’re now halfway through this comparison of Azure Lambda, next up is deployment.

For AWS Lambda, perhaps the most popular deployment method is the Serverless command-line tool. Alternatively, you can upload a zip folder on the AWS Lambda console. For Azure Functions, the most popular deployment method is the VS Code extension. You can also upload a .zip file using a REST API or Powershell. See this for more information.

Concurrency

AWS Lambda has a default concurrency limit of 1000 invocations per region. However, you can ask for a limit increase by contacting AWS Support. Azure functions can scale out to a maximum of 200 instances on Windows and 100 instances on Linux.

Exclusivity

Some applications leave you with no choice but to use one out of the two. For example, if you wish to create a Smart Home skill on Alexa, then your Cloud endpoint has to be an AWS Lambda. It simply does not accept any other Serverless function endpoint. Similarly, on certain AWS services (like Cognito or IoT Core), if you wish to run a custom authentication routine, you need to use AWS Lambda only.

Code Reusability and Isolation

Here’s the last parameter of Azure Lambda comparison, code reusability & isolation. AWS Lambda has the concept of layers that can be used for isolating your application code from the rest of the code (libraries, packages, etc.). These layers can be reused across lambda functions, even across accounts. Thus, if a number of Node.js Lambda functions, across services, use the same Node.js packages, you can add the packages in a layer and reuse the layer across all the functions. Note that a maximum of 5 layers are allowed per Lambda function, and the total size of the application code + layers should be less than 250 MB.

In the case of Azure functions, all functions within a Function App can use the same packages. However, different Function Apps cannot reuse the packages, and separate deployments have to be made for separate Function Apps. However, there is no limit to the size of the deployment package. You will be charged a higher storage cost for a higher deployment package size though.

This brings us to the end of the Azure Lambda comparison.

Conclusion

This article discussed how Azure Lambda functions compare on several parameters. Whether to use AWS Lambda or Azure functions depends a lot on your application and constraints. In many cases, mere familiarity with one over the other drives the decision. Though not exhaustive, attempts have been made to cover the most commonly considered parameters in this article, and, hopefully, it will serve as a guide if you are genuinely evaluating the two. Thanks for reading.

However, it’s easy to become lost in a blend of data from multiple sources. Imagine trying to make heads or tails of vague sets of data. This is where Hevo comes in.

visit our website to explore hevo

Hevo Data with its strong integration with 100+ Sources allows you to not only export data from multiple sources & load data to the destinations, but also transform & enrich your data, & make it analysis-ready so that you can focus only on your key business needs and perform insightful analysis.

Give Hevo Data a try and sign up for a 14-day free trial today. Hevo offers plans & pricing for different use cases and business needs, check them out!

Share your experience of learning about Azure Lambda functions in the comments section below.