Topic 44

Managed vs Do-It-Yourself

Concept

Every time you use a cloud service, the same quiet decision has already been made: someone chose how much of the underlying work to hand to the provider and how much to keep in-house. This choice appears over and over — for databases, for scaling, for security, for almost everything.

It has a name: the managed versus do-it-yourself trade. Understanding it gives you a mental shortcut that cuts through most cloud conversations.

Think of it like housing. A furnished flat with utilities included (managed) means you pay rent, move in, and the landlord handles the boiler, the roof, and the wiring. Buying a house (do-it-yourself) gives you more say over everything — but your weekends disappear into maintenance. More rent versus more of your time. That's the trade.

The Same Pattern, Everywhere

For nearly every cloud capability, there are two options: a managed service, where the provider installs, patches, backs up, and scales the thing for you; and a DIY (do-it-yourself) option, where you rent raw machines and install and run it yourself. A managed database is an example of the first: the provider handles backups and updates. Installing a database on a virtual machine you control is the second.

What Is "Undifferentiated Heavy Lifting"?

Cloud teams use the phrase undifferentiated heavy lifting to describe the unglamorous work that every team has to do but that doesn't make anyone's product special. Applying security patches. Performing nightly backups. Scaling capacity when traffic spikes. Replacing a failed disk.

None of those tasks appear in your product's feature list. Customers don't pay for "well-patched operating system." When a managed service handles that work, your team stops spending time on it — not because it disappears, but because someone else does it.

The Trade — What You Give and Get

Managed services cost more in dollars. DIY costs more in time, expertise, and risk. Neither is universally right. The question is always: is this the kind of work your team should be spending its time on?

A team whose product is a database might have good reasons to run their own. A team whose product is a photo-sharing app almost certainly does not need to become experts in database internals. They should pay for managed and get on with building the app.

The Usual Default

Most experienced cloud teams apply a simple rule: prefer managed for anything that isn't your core product's special sauce. Use a managed database, a managed message queue, a managed caching layer. Keep DIY only for the things that directly define what your product does differently. This isn't laziness — it's focus.

Managed or DIY? A decision guide
Is this capability your product's core differentiator?DIY — control matters
Is it important plumbing that isn't your unique value?Managed — let the provider run it
Does your team have deep expertise in running this?DIY may be viable — weigh the cost
Would a failure here be hard to recover from?Managed — providers have dedicated reliability teams
Three cloudsAWS pushes managed hard; hundreds of managed services across every categoryGoogle Cloud same push; managed defaults throughoutAzure same approach; the managed vs DIY trade applies identically across all three
Common Confusions
  • "DIY is always cheaper." Cheaper in dollars, yes — but far more expensive in engineer time, on-call burden, and the risk of getting things wrong. Total cost includes both.
  • "Managed means losing control of your product." You control your application and your data. The provider only runs the infrastructure underneath. Your business logic stays yours.
  • "Real engineers always self-host everything." Experienced teams deliberately choose managed for non-core work. Self-hosting everything is a sign of misplaced priorities, not engineering skill.
  • "This choice is made once and stays fixed." Teams regularly migrate from DIY to managed (or the reverse) as their needs evolve. It's a recurring decision, not a one-time commitment.
Why It Matters
  • This mental model shows up in nearly every cloud decision — recognizing the pattern lets you follow (and contribute to) those discussions immediately.
  • Understanding "undifferentiated heavy lifting" explains why managed services exist and why teams pay for them even when a DIY option is technically available.
  • For managers, it's the clearest way to frame cloud spending: you're buying your team's time back from the plumbing so they can build what actually matters.

Knowledge Check

What does "undifferentiated heavy lifting" mean in cloud terms?

  • Necessary maintenance work — patching, backups, scaling — that every team must do but that doesn't differentiate any product
  • Computationally intensive tasks that require large, expensive machines to run
  • Physical work like installing servers in a data center by hand
  • A type of cloud pricing where a flat fee covers all infrastructure costs">A type of cloud pricing where a flat fee covers all infrastructure costs

When should a team generally prefer a managed service?

  • Only when the team is too small to have any dedicated engineers
  • When the capability isn't the product's core differentiator and the team's time is better spent elsewhere
  • Whenever the managed option costs less money than running it yourself
  • Only for databases, never for networking or computing">Only for databases, never for networking or computing

With a managed service, what do you keep control of?

  • The operating system and its security patches
  • The physical hardware the service runs on
  • Your application logic and your data
  • Security patches to the underlying database

You got correct