Vagrant vs. Terraform: an alternative interpretation
Preface
I was curious about the differences between Vagrant and Terraform. When would someone choose or prefer one technology over the other?
At their core, both are designed to automate the creation of virtual machines and the the recreation of computing environments. Pretty much the Docker Compose of virtual machines.
So like everyone, I googled and found an authoritative article from their creator and maintainer HashiCorp aptly titled “Vagrant vs. Terraform”. I read – and had to re-read a couple of times – nitpicking over the choice of words in order to spot the subtle differences and implications behind the jargons and terminologies.
Criticisms of HashiCorp’s explanation
Why is it difficult to understand you may ask. Let me quote a few examples.
Example 1
Vagrant is a tool focused for managing development environments and
Terraform is a tool for building infrastructure.
Isn’t creating and recreating a development or test environment “building infrastructure” too?
Example 2
Terraform can describe complex sets of infrastructure that exist locally
or remotely.
Given a long enough Vagrantfile written in Ruby, perhaps it can also recreate
a similarly complex setup with multiple virtual machines? So what are the things
it cannot do that Terraform can? This does not elaborate further in details.
As a side note, I suspect the network configuration options in Vagrant are more limited and less granular than those available in Terraform which allow users to explicitly describe complex networking between virtual machines in different cloud service providers, though this needs to be confirmed.
Example 3
Terraform is focused on infrastructure management and not development environments
People may be wondering how different a production environment and development environment can be. In some cases, developers try to recreate the development environment as closely as possible to the real thing, to the point of creating an exact duplicate clone setup. It all depends on the simplified adaptations adopted for a development setup. And these differ between every organization, system and project.
For users whose development, testing, staging and production environments are exact replicas or clones of one another, would it make a difference between choosing Vagrant or Terraform? And what are the implications for choosing one over the other?
These are unanswered.
Example 4
Vagrant provides a number of higher level features that Terraform doesn’t. Synced folders, automatic networking, HTTP tunneling, and more are features provided by Vagrant to ease development environment usage.
This is because it is assumed that the virtual machines are running locally on one single development machine host, which allows all these automated network configuration and storage volume/network shares mounting (synced folders). The assumption is not stated.
Example 5
Finally, it finishes off by concluding that:
The primary usage of Terraform is for managing remote resources in cloud providers such as AWS. Terraform is designed to be able to manage extremely large infrastructures that span multiple cloud providers. Vagrant is designed primarily for local development environments that use only a handful of virtual machines at most.
Vagrant is for development environments. Terraform is for more general infrastructure management.
The final two paragraphs tell us what we actually need to know, and I agree this main message is helpful: Terraform is for large production cloud/remote environment, and Vagrant is for small local development environment.
Still, people are left to wonder how “extremely large” of an infrastructure it should be to definitely choose Terraform over Vagrant. And what do we mean by “a handful”? Two, five or ten VMs? And is managing development environment part of infrastructure management? The clarity is not there.
Reading between the lines
My conclusion: it is not easy to read and understand for anyone new to virtual machines, cloud computing and business users. The differences are not obvious and we have to read between the lines carefully despite HashiCorp’s attempt at explaining the differences to help people decide which one is for them.
Whether it is carefully worded to obfuscate the differences or it happens to be a lame attempt at clarifying the differences, I certainly hope that HashiCorp can improve its documentations and communication with its customers and users.
A clear explanation just needs one line or paragraph that summarizes the core information to help people make a decision which one is for them.
This blog post is my own interpretation to help people interested in Vagrant and Terraform to decide which one is more suitable for their use cases.
The Real Differences
In short, the differences are:
| Aspect | Vagrant | Terraform |
|---|---|---|
| Year | 2012 | 2014 |
| Language | Ruby | YAML |
| Targeted user | Developers | Deployment / operations engineers |
| Use case | Creating development enviroment on a single local machine | Creating production and test environments having virtual machines potentially spanning multiple cloud and on-premise servers |
| Host Virtual Machine | Local developer workstation or desktop PC | Cloud service providers, remote data centers or on-premise servers |
| Hypervisor | Type 2 (e.g. Oracle VirtualBox, VMWare Fusion/Workstation, Microsoft Virtual PC, Parallel Desktop) | Type 1 (e.g. VMWare vSphere/ESXi, Microsoft Hyper-V, KVM, Xen) |
Note that this is a simplication to compare and contrast to help people make a decision.
Disclaimer for edge and atypical use cases
Certainly, there could be cases where people install a type 1 hypervisor on a workstation or desktop PC for small scale experimentation or repurposing consumer grade hardware for a home server. And this opens up the door to test- drive Terraform on a local machine.
And there could be people deploying a simple setup consisting of one virtual machine instance to a public cloud provider like Amazon, Azure or Google Cloud Platform. In this case, Vagrant can possibly achieve this too with the right plug-in, as well as Terraform.
Or at the other extreme end of the spectrum, people writing long elaborate
Vagrantfile just to mimic a very complex setup running on a type 2 hypervisor
on a single host. That single virtual machine host may well be a souped-up
machine with dual Intel Xeon or AMD EPYC processors paired with 128 GB of RAM,
enough to run quite a few virtual machines concurrently. Or some old rack or
blade server lying on someone’s table?
Who can predict what creative people would attempt to do with what they have or can afford?
And then, it may boil down to user’s preferred language (Ruby or YAML) to describe their virtual machine environment.
Conclusion
Vagrant and Terraform are both good products in their own right, and each has their own place. It is just a pity that HashiCorp has not made it easy for people to decide at one glance which one is more suitable for them.
I understand that Vagrant is an open source community-driven technology that HashiCorp does not make money from. Instead it has the incentive to push for and promote Terraform which actually brings it money. This may be the business reason behind its less proactive attitude toward it.
Hopefully, my own overly-simplified (and perhaps less accurate) personal interpretation can help some people to decide which technology is more suitable for them to investigate for learning and adoption.