Clarity on Azure DevOps parallel job "consumption"

Question

We're in the process of moving away from Visual Studio's web publish into Azure DevOps and require some clarity on how parallel jobs are "consumed", or an alternative way to quickly deploy our application to multiple destinations.

Our web application has a single code-base and is published into 15 separate clients' hosting areas (all on Azure App Services), each with their own configuration file transform during the final deployment process, e.g.:

Currently the process is working well on Azure DevOps' "free" account, but it takes around one hour to deploy to each client hosting environment as the processes run in a queue (Deploy Customer 1 → Deploy Customer 2 → ...).

Yesterday to streamline the process and get the deployments done asynchronously we purchased three "Parallel Jobs" from the billing area. Deployment was indeed much quicker as three ran in parallel. However, when they were finished there were no parallel jobs available in our account.

The terminology in the Learn portal talks about monthly resources, so we had assumed that the parallel job agents were a one-month purchase (buy once/use many in that month).

If they are consumed immediately upon completion then according to the pricing calculator, if have a weekly release and use 15 parallel jobs then that will cost me over £2,000 per month (£35.34 * 15 * 4) which seems unbelievably expensive for a short time saver.

Can anyone please clarify the situation, or suggest an alternative method for deploying asynchronously?

Answer 1

I can't comment, so using an answer.

What exactly do you mean by "However, when they were finished there were no parallel jobs available in our account." If you use the MS hosted agents, and if you bought three parallel jobs, you should see them in the Organization settings -> Parallel jobs overview. If not, try lo log out, clear cache etc. The purchases are indeed for a month, as explained here, the costs would be about 40 euros for a month, so for three it would be around 120 euro a month.

As an alternative, and as suggested by others, you could use self hosted agents. They are cheaper, and depending on resources on the host where you install the agent, they run faster.

And then, depending on your needs, there are several ways that might speed up your deployment process, but that usually depends on what you've already configured and your needs.

And then last, if you are a really early user of DevOps, there was an option to pay for hosted agents by the minute. That could explain your situation, but I've checked several organizations inclusing a few already several years old but I can't find any reference of it.

Answer 2

If your pipeline exceeds the maximum job timeout, try splitting your pipeline into multiple jobs. Here is the docs for it https://learn.microsoft.com/en-us/azure/devops/pipelines/process/phases?view=azure-devops

Answer 3

So we use privately hosted agents and it saves us a lot of money, here are some details

Public project Up to 10 free Microsoft-hosted parallel jobs that can run for up to 360 minutes (6 hours) each time No overall time limit per month

Private project One free job that can run for up to 60 minutes each time 1,800 minutes (30 hours) per mont. You can pay extra to get parallel jobs.

Self Hosted Agent You can just setup your own self hosted agents, it just a script you run on a VM, takes about 10min per agent plus of course you need to install the software you need on the agent(s) and then you don't have those restrictions.