What we do
Tools
Industries
Department
How we workLet's talk →
Home/Database Developer
Database Development

Hiring a Database Developer: Slow Queries, Risky Migrations and Untested Backups

Most software mistakes can be fixed in the next release. Data mistakes often can't: a migration that drops rows, a script that overwrites the wrong column, a backup that turns out to be empty on the day you need it. That's why database work deserves a slower, more careful hire than most, and why it helps to know which of the usual problems you actually have.

Four questions about your database to answer before you hire

Each answer points to a different kind of database job, and none of them needs a developer to find out.

  • When did someone last restore a backup, not just take one?

    If the answer is never, you don't know whether the backups work or how long getting back would take. That's the first job, before any other change.

  • Which screens or reports got slower this year, and roughly when?

    A slowdown that tracks business growth points to tables that outgrew their indexes. A sudden one points to a release or an import that changed something.

  • Do the same customer, product or job show up twice with slightly different details?

    Duplicates usually mean the database isn't enforcing the rules the business assumes, so every new connection or import adds more.

  • Is something moving soon: a new system, a new host or the cloud?

    A migration is the riskiest single event in a database's life. It needs a rehearsal and a way back, planned before the date is set.

Why a query that was fast last year is slow now

The database usually gets the blame. Often it's the way it's being asked.

An index that fit a small table no longer fits the questions

How you'll know: Searches and reports that filter by date, status or customer slow down steadily as the business grows.

Fix: Read the query plan (EXPLAIN, or the execution plan on SQL Server) for the slow queries. A full scan of a large table where you expected an index lookup shows exactly which index is missing.

The app asks a thousand small questions instead of one

How you'll know: A single page triggers hundreds of near-identical queries, each fast on its own.

Fix: This is the N+1 pattern that ORMs make easy to write by accident. It's fixed in the application code by loading related records in one query, which is why a database developer should read the app too, not only the database.

A calculation on a column hides it from its index

How you'll know: A query filtering on something like the year of an order date is slow even though the date column is indexed.

Fix: Rewrite the filter as a range on the raw column, such as dates between two values, so the index can be used.

Long jobs lock the tables everyone else needs

How you'll know: Everything stalls at the same time each day, often during an import, an export or a nightly sync.

Fix: Split big updates into smaller batches, move heavy reporting to a read replica or a copy, and schedule the rest away from working hours.

A database migration with a way back at every step

Whether it's a new schema, a new engine or a move to the cloud, the order is much the same.

  1. 1

    Take a backup and prove it restores

    Restore it somewhere else and open the data. That's your way back if everything after this goes wrong.

  2. 2

    Write the migration as scripts, kept in version control

    Schema changes and data moves as files someone can read, review and run again, not clicks in an admin tool that nobody can repeat.

  3. 3

    Rehearse on a full copy and compare

    Run the scripts against a copy of the real data. Compare row counts table by table, and totals that matter to the business, like invoice sums or stock on hand, between old and new.

    Watch for: rows that fail quietly: dates in odd formats, text in the wrong character set, records that point at something deleted years ago.

  4. 4

    Add before you remove

    For changes to a live system, add the new column or table first, fill it, switch the app to read from it, and only drop the old one later. That's how most migrations run without downtime.

  5. 5

    Cut over with a time limit and a rollback plan

    Agree in advance how long the switch may take and what result means going back. When the comparison checks pass on the live run, the old system can retire.

Let's talk

Your data, moved and checked by more than one pair of eyes

n-frames handles database work for businesses whose orders, jobs and customer records live in it: finding why queries slowed down, adding the constraints that stop duplicates and broken links at the source, testing that backups really restore, and running migrations that are rehearsed on a copy and reviewed by a second engineer before any live data moves. When the database sits behind an app or an integration, the backend developer and API developer pages cover those sides, and hosting and deployment are on DevOps engineer.

Rehearse our migration on a copy →

Questions people ask

Can you migrate our data without downtime?

Often, using the add-before-remove approach above, depending on the engine, the size of the data and how many systems write to it. Some moves do need a short window. If yours does, you'll know how long and why before the date is set.

Is a database developer different from hiring a backend developer?

Yes, in focus. A backend developer builds the server side that uses the database. Database work is about the data itself: how it's structured, how fast it answers, the rules that keep it correct, and moving it safely. Many projects need both, and n-frames covers both.

What if we're not sure the database is actually the bottleneck?

That's common, and finding out comes first. We time the slow screens, look at what queries they send and read the query plans. Sometimes the fix is an index, sometimes it's the app's code, and sometimes it's the server it runs on.

How do we know our backups actually work?

Only by restoring one. Restore a recent backup onto a separate server, check the newest records are there, and time it. Do it again after any big change to the database or the hosting.

We've outgrown a spreadsheet or an Access database. Is that database work?

Yes, and it's one of the most common versions of it. The data moves into a proper database with rules that stop the duplicates and blanks a spreadsheet allows, usually with simple screens on top so the team keeps working the way it's used to. If the data sits on a cloud host, AWS developer covers that side.

Let's talk

Tell us the one thing your team does manually that eats up time. We read every message and reply within a day.