Scheduled jobs are common in software systems, most of them perform some kind of a network or data related activity, outside the main execution context and triggered based on a timer event.

There are several options available in Azure to implement scheduled jobs.

1. Serverless options
2. Azure Scheduler
3. Azure Batch
4. Custom code in VMs

Before digging into the details, first rule of design in my space, I always advocate the developers not use scheduled events in any case that you can trigger it using any other trigger. I agree that in certain cases the scheduled jobs cannot be avoided like a nightly data loading job from an OLTP to a data warehouse or your nightly builds but checking a folder for a file can be triggered using a different event rather than checking it in intervals.

Now let’s see the different available options.

Serverless

Azure has two different serverless offerings, Azure Functions and Logic Apps. Both can handle timer based triggers.

Azure Functions are straight forward, you can specify a cron expression on the timer trigger and write your code and deploy. Done! You can do this either from Visual Studio or in the ‘integrate’ section of your function in the portal. In the code it is set as follows.

public static class Function1
{
[FunctionName("Function1")]
public static void Run([TimerTrigger("0 */5 * * * *")]TimerInfo myTimer, TraceWriter log)
{
log.Info($"C# Timer trigger function executed at: {DateTime.Now}");
}
}

Read more about cron expressions here

We can configure the scheduled jobs in Azure Logic Apps using the recurrence trigger as shown below. Logic Apps designer is straight forward and we can configure the recurrence timer trigger and schedule the following actions.

Azure Scheduler

Azure Scheduler is a PaaS based scheduler.  This is a packaged consumption model of the scheduled tasks, but this has limited actions.

You can use HTTP/HTTPS calls or put a message in a Storage Queue, or put a message in Service Bus topic/queue.  The HTTP/HTTPS can be used to communicate with external APIs in a schedule.  The serverless triggers greatly make the Azure Scheduler obsolete, the recurrence logic has good customization out of the box, but again the functionality can be easily added in the serverless model.

Azure Scheduler had been used heavily before the serverless options were available. Especially PaaS developers used this approach in order to trigger timer actions by simply using HTTP/HTTPS triggers and hit backend services to execute the logic. Now, this is a high time to consider the rewrite of those schedule jobs using serverless model.

Azure Batch

Azure Batch provides the infrastructure for the batch processing. Azure offers this to use cheap computation for the batch processing jobs and also for computation heavy processing. Though the main intention of Azure Batch is not running a scheduled task in for an application, the application specific jobs can be submitted to the Azure Batch and executed on a schedule.

Example, in a HR application running on Web Apps require to process large amounts of data at the end of every month to calculate the wages could be a Azure  Batch task which runs every month.

Custom code in VMs

This is the traditional approach. If you’re using IaaS then this option will not make you feel outdated. Also, in certain cases you would need this especially when your scheduled job requires some libraries or frameworks which require VMs.  You can write the code and execute them as cron jobs in Linux VMs or as a process using Task Scheduler in Windows VMs. In Windows you can also use Windows Services for the schedule jobs.

Do not use Windows services unless you require your logic to be executed outside the system context. Most of the scheduled jobs can be run as tasks using Task Scheduler.