The New Way to Scale

Scalability is one of those poorly understood concepts in the computer world, and it is understood even worse in the business world.  Briefly, it is the ability of a system’s capacity to grow by adding more resources to it.  A desirable property is that growing the capacity by a unit should never require a single huge investment.  If it is the case that each additional unit of capacity requires a constant amount of additional resources, we say the system scales linearly.

Most companies today do not scale

Think about car assembly lines: they represent a huge infrastructure investment of hundreds of millions of dollars, optimized to produce a single model of car at a particular rate of completion.  The line is most efficient when working 24×7, producing a steady stream of new cars.  If the demand does not meet that capacity, then the margin per unit is less than planned, and possibly the car is not viable.  The company’s response is to shutter the plant and throw everyone out of work.

On the other hand, if the demand is greater than the planned capacity, there is no way for the company to meet that demand without taking the risk that the second plant may never run at full capacity, and the product will not be viable.  This risk is especially strong for your typical fad product – the wildly successful Tickle-me-Elmos whose demand outstrip supply for a few short weeks.  Because the manufacturing takes place on the other side of the planet, in a factory of fixed capacity, there is no way to get more Elmos on a ship before Christmas, let alone in stores.  They arrive, instead, after their moment and we get an Elmo glut in January.

Actually, I have no idea if Elmo production failed to scale, but certainly the weeks it takes to ship from China mean that this type of manufacturing fails to scale quickly.

When we scale computer systems, we talk about scaling out or scaling up.  To simplify, scaling up means you buy a bigger computer; if your single processor machine no longer serves, you get a four-processor machine, and so on.  The problem with this strategy is that the first computer is now surplus, and the investment to scale bigger and bigger gets bigger.  Ultimately, you hit a brick wall, and nobody can supply enough processors in a single frame for your needs (that numbers crazy high today, but CPU is only one resource, and other resources hit the wall earlier).

Scaling out is generally favored over scaling up.  When we scale out a system, we start with one low-powered machine, and when demand outstrips its capacity, we add a second identical machine, then a third and so on.   Theoretically, this can go on forever, although practically, you hit design limits on a resource that only scales up; database servers, for example, are often designed to scale up because, well, it’s hard to come up with a database architecture that scales out.

How can we take the idea of scaling out to manufacturing, and how does the small batch platform enable it?

Suppose you are a small-batch manufacturer, creating Elmo dolls and selling them into your community.  Suddenly they become wildly popular.  Your neighbours start complaining about the lineup out the door of your small shop; you’re working all night to craft more Elmos on your single set of machines, and they are snapped up each day before noon.  Customers camp out the night before to ensure being first in line for Elmo in the morning and your neighbours complain even more.

You need a way to scale.  Fortunately, your manufacturing platform consists of flexible machines – CNC mills and lathes, 3D printers, laser cutters, and the like.  So, production is easy to replicate.  You could set up a bigger factory, but there’s no room in your small shop, and so, that would take time.

What you need to do is find a partner with a similar capability and license them to produce and sell your Elmo.  They do this in their own community, getting their own lineup of customers out the door, and pay you a license fee for your design.  Both of you are happy.

Not only are you happy, but your customers are happier.  They no longer have to line up outside your small shop to get an Elmo, but can visit your partner whose shop is closer to them.  The planet benefits from this proximity.

What’s more, you and your partners are ramping Elmo production up and down as demand increases and ultimately wanes.  This means there is no post-yuletide Elmo glut, and again the planet is happier.

Scaling up production by scaling out, and then scaling down again is one of the ways that new companies will be more efficient than old-style manufacturing.  And that is one of the reasons why new-style companies are going to eat old-style companies for lunch.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: