Trading platforms, charting tools and portfolio dashboards all run on servers somebody has to look after. When those servers are set up by hand, the setup lives in a few heads and a pile of screenshots. A newer habit, often labelled InfraOps, replaces that with written definitions that can be read, reviewed and replayed. This explainer uses a household-ledger comparison to make the idea plain. It is general technology education only and offers no investment, tax or financial advice.
Start with a kitchen-table ledger
Imagine a family that tracks money with three records. There is the notebook where they write down what they plan to spend, the bank statement that shows what actually left the account, and the budgeting app’s own tally of what it believes happened. When all three agree, life is calm. When they do not, the argument is never really about the money; it is about which record to trust.
Infrastructure has exactly the same trio. The Ops Atlas that xOps School maintains calls them code, state and reality. Code is the written intent, state is the tool’s own tally of what it built, and reality is what the cloud provider is really running. The page that sets this out, the Atlas explainer on InfraOps, also lists four ways they drift apart, and each has a household equivalent.
Four ways the records disagree
- Written but never done. You noted a payment in the notebook, but it never happened. In infrastructure, a change was approved in the code but never applied.
- Changed behind your back. The bank statement shows a charge you did not plan. In infrastructure, somebody altered a live setting by hand, which engineers call drift.
- Never recorded. A standing order from years ago that none of your records mention. In infrastructure, a resource created manually and never described anywhere.
- Recorded but gone. The app lists an account you closed. In infrastructure, the tool’s tally mentions something that has already been deleted elsewhere.
Most real trouble comes from the third item. A tidy report from the budgeting app tells you nothing about a standing order it was never told about. In the same way, a clean preview from infrastructure tooling tells you nothing about unregistered resources.
Why a “preview” is such a big deal
Before the tools make a change, they can print a list of what they intend to add, modify or remove. Think of a bank letting you see tomorrow’s scheduled payments before they go out. Reading that list carefully is the single cheapest safety measure in the field. It is also perishable: should one colleague apply a change first, any list a second colleague drew up against the same starting point is now stale, just as a payment schedule you checked on Monday may be different by Wednesday.
What a rollback is not
In a bank, a failed transfer is usually reversed automatically. In infrastructure tooling, a change that fails halfway typically does not roll back. You end up with a half-finished state and fix it by preparing a new change that goes forward. That is why reviewers look so hard at previews before anything runs, and why nobody should ever force open a lock that another person took.
Secrets belong in a locked drawer
A ledger sometimes carries account numbers. Infrastructure records are similar: state files and saved previews can hold sensitive values. The Atlas notes that newer features exist to keep secrets out of stored records, and that even the credentials used for previews should be treated with serious care. If you are using a hosted trading or analytics product, this is the kind of question you can reasonably expect its provider to answer in plain words.
Counting the right things
Good budgeters do not boast about how many categories they track; they check whether the totals reconcile. The Atlas suggests similar honest measures for infrastructure: what share of everything in scope is covered by written definitions, how often reality drifts, how long a difference stays open, and how old the server images are. Its worked example shows 1,512 managed items against 1,840 in scope, giving 82.2 percent coverage, then insists on restating it as 328 items nobody reviews. That reframing is the same discipline as reading the unexplained line on a statement instead of admiring the balance.
Three small misunderstandings to avoid
First, InfraOps is not software you install. In the Atlas it is a casual label that no standards organisation backs, applied to the lifecycle of infrastructure that is written as code. Second, it is not a promise of safety. Written definitions can contain mistakes just as a ledger can contain a typo; the gain is that mistakes become visible and reviewable. Third, it does not replace judgement. A preview shows what will change, not whether the change is wise.
If you remember only one comparison, remember the bank statement. It is the record that the outside world keeps about you, and it does not care what your notebook says. Infrastructure has that same stubborn outside record, the provider’s, and every good habit in this field is a way of checking your own notes against it regularly.
Why a markets site cares
Readers who follow markets rely on data feeds, alerts and dashboards that must be up when it matters. Those who build or run such tools, even as a hobby project, benefit from being able to rebuild a setup from written definitions rather than from memory. Our own pieces on the human side of trading, such as a step-by-step guide to keeping your trading routine disciplined and a look at the mindset habits that experienced traders describe, make a parallel point: process and records beat improvisation.
A practical, small start
- Write down the pieces of your setup in a plain file, even if it is only a hobby dashboard.
- Note where each secret lives and who can read it.
- Keep a running log that notes the date and the reason for every edit.
- Once a month, compare your notes with what is actually running, and fix any difference.
- Try rebuilding the whole thing once, to learn what your notes forgot.
None of this predicts anything about markets or tells you what to buy or sell. It simply keeps the machinery honest, so that your attention can stay on your own decisions. As with a household ledger, the aim is boring accuracy, and boring accuracy is a very respectable goal.