The History of Pets vs Cattle and How to Use the Analogy Properly | Cloudscaling

The History of Pets vs Cattle and How to Use the Analogy Properly | Cloudscaling ">

I have been meaning to write this post for a long time, but one thing or another has gotten in the way. It’s important to me to provide an accurate history, definition, and proper usage of the Pets vs Cattle meme so that everyone can understand why it was successful and how it’s still vital as a tool for driving understanding of cloud. The meme has taken off because it helped created an understanding of the “old way” vs. the “new way” of doing things. That’s great, but the value of the meme becomes muddied when misused. We can all agree there’s enough muddy terminology and phraseology already, such as “cloud,” “hybrid,” and “DevOps”. So this post aims to set the record straight and assure a canonical history that everyone can reference and use.

The History

Some time in 2011 or 2012 I was struggling with explaining to customers how AWS, cloud native apps, and cloud more generally was fundamentally different from what had gone before[1]. Most explanations required an exorbitant amount of time. I was searching for a way to synopsize this when I came across a great little presentation from Bill Baker about Scaling SQL Server. Bill wasn’t running around presenting this in cloud circles. I dug it up through Google searches. Bill wasn’t talking about cloud or cloud computing. His focus was about “scale-up” vs. “scale-out” architectures generally.

Most importantly, Bill used an analogy that resonated deeply with me. He talked about scale-up vs. scale-out in the context of comparing a pet and a herd of cattle. A light went on above my head and I realized this was it, the elevator pitch I was looking for. But there was a slight, but important nuance that I focused on. First, I put pets vs cattle in the context of cloud, and second I emphasized the disposability of cattle and the uniqueness of pets. This was much more important than whether you scale-up or scale-out. Those are, I believe, side effects, of how you view a server. If you view a server (whether metal, virtualized, or containerized) as inherently something that can be destroyed and replaced at any time, then it’s a member of the herd. If, however, you view a server (or a pair of servers attempting to appear as a single unit) as indispensable, then it’s a pet.

Here is the elevator pitch I would give in presentations, pretty much verbatim:

In the old way of doing things, we treat our servers like pets, for example Bob the mail server. If Bob goes down, it’s all hands on deck. The CEO can’t get his email and it’s the end of the world. In the new way, servers are numbered, like cattle in a herd. For example, www001 to www100. When one server goes down, it’s taken out back, shot, and replaced on the line.

This explanation of Pets vs Cattle is what resonated with Tim Bell at CERN and many others and caused them to replicate and propagate the analogy, which created the meme that has edified so many and has so cleanly represented the transition we are all going through to cloud.

Understanding Pets and Cattle

Let’s take a minute to clearly define pets and cattle.

Pets

Servers or server pairs that are treated as indispensable or unique systems that can never be down. Typically they are manually built, managed, and “hand fed”. Examples include mainframes, solitary servers, HA loadbalancers/firewalls (active/active or active/passive), database systems designed as master/slave (active/passive), and so on.

Cattle

Arrays of more than two servers, that are built using automated tools, and are designed for failure, where no one, two, or even three servers are irreplaceable. Typically, during failure events no human intervention is required as the array exhibits attributes of “routing around failures” by restarting failed servers or replicating data through strategies like triple replication or erasure coding. Examples include web server arrays, multi-master datastores such as Cassandra clusters, multiple racks of gear put together in clusters, and just about anything that is load-balanced and multi-master.

The key here is that in the old world redundancy through having two of everything, the ubiquitous HA pair in the enterprise datacenter, is not enough. What is required is assuming that failures can and will happen. That every server, every component is able to fail without impacting the system.

Most importantly, this analogy, as presented above, has helped educate a generation of IT managers, CIOs, and others, while giving them a tool for further explaining the old vs new.

Going Off The Ranch

That’s why it’s important to stick to or, at least start with, the core message above. People can and will, take this simple analogy and turn it to their own uses. Folks have added to the analogy (e.g. “pets, cattle, & ants”, “children, pets, and cattle”, etc.) and have modified it in various manners....