StormaticsStormatics

Choosing a PostgreSQL DBA Service

A PostgreSQL DBA service is only as good as the worst night your database has. The sales deck won’t tell you how a provider behaves at 3 AM on a Sunday, but the right questions will.

You’ll see these services under a lot of names: remote DBA, outsourced DBA, managed DBA services, PostgreSQL DBA support. At Stormatics we call it DBA as a Service. The idea is the same. Instead of hiring a full-time DBA, you get a team of PostgreSQL experts who run your databases under an SLA.

I’ve spent most of my career on both sides of this, running PostgreSQL teams and buying services for them. Here’s what I’d look for if I were choosing a provider today.

What a remote DBA service must include

Every provider says “proactive” and “24×7”. These are the things that tell you whether they mean it.

  1. A written SLA with response times by severity. Not “best effort”. You want a number for a production-down incident, a number for degraded performance, and a clear definition of each. Ask whether the clock starts when you open a ticket or when someone picks it up.
  2. Coverage that matches your business. If your customers are awake around the clock, business-hours coverage leaves you alone for 128 of the 168 hours in a week. Know which hours you’re buying.
  3. Tested restores, not just backups. A backup you’ve never restored is a hope, not a plan. A good service restores your backups on a schedule and tells you how long it took, so you know your real recovery time.
  4. Upgrades as part of the job. PostgreSQL ships a major version every year and supports each one for five years. Your provider should plan minor updates routinely and major upgrades well before end of life, with minimal downtime.
  5. Monitoring that a human acts on. Alerts on replication lag, bloat, connections, disk and long-running queries are table stakes. What matters is who reads them and what they do next.
  6. PostgreSQL support, wherever it runs. You shouldn’t have to move to a vendor’s fork or a specific cloud to get help. Good providers are platform agnostic: on-premises, any cloud, Kubernetes or a managed service like RDS.
  7. A regular health report. Monthly or quarterly, in plain language: what changed, what’s at risk, what they recommend. This is how you know the work is happening when nothing is on fire.

Questions to ask a provider

The answers here tell you more than any feature list.

  • Who answers at 3 AM, and how senior are they? A first-line ticket queue that escalates to an expert an hour later isn’t 24×7 expert support.
  • Will we have named engineers who know our setup? Context saves the first 30 minutes of every incident.
  • Walk me through your last production-down incident. Listen for the timeline, the root cause and what changed afterwards.
  • How do you onboard a new customer? The right answer starts with an assessment of your current setup, not with a monitoring agent install.
  • What isn’t included? Every service has limits. A provider who explains them up front is one you can trust when a bill arrives.
  • What happens if we leave? Your runbooks, configs and documentation should be yours.
  • Do your engineers contribute to PostgreSQL? Not a must, but people who work on PostgreSQL itself tend to diagnose it faster.

Red flags

If you see these, keep looking.

  • No published or written response times. If they won’t commit to a number, they won’t meet one.
  • Lock-in to a proprietary fork or tool before they’ll support you.
  • Hourly billing for emergencies with no cap. You’ll hesitate to call them, and that’s the worst habit to build during an outage.
  • No onboarding assessment. A provider who starts managing your database without a health check is learning your system during your first incident.
  • “We’ve never lost data” with no description of how restores are tested.
  • Juniors on the contract, seniors in the sales call. Ask who will actually do the work.

How pricing models compare

Most PostgreSQL DBA services price in one of four ways, and per core is now the most common. None is wrong; each one fits a different shape of estate.

ModelHow it worksWorks well whenWatch out for
Per coreA fee for each CPU core (vCPU) on the servers PostgreSQL runs onYour instances vary in size, and you want cost to follow the capacity you runHow replica and DR cores are counted, and the bill when you scale up for a peak
Per instanceA fixed yearly or monthly fee for each PostgreSQL instance under managementYou want a predictable budget and have a stable number of instancesHow replicas are counted, and minimum instance counts
Per hourYou buy a block of hours, used for routine work and incidentsYour needs are occasional or project shapedIncidents eating the block, and slower response when hours run low
Per clusterOne fee for a primary and its replicasYou run large HA clusters with many replicasWhat “cluster” means when you add read replicas or a DR site

Whichever model you’re quoted, turn it into a yearly cost for your whole estate and put it next to the cost of hiring. One fully loaded senior PostgreSQL DBA in the US costs $153K to $198K a year. For true 24×7 coverage you need at least three, which puts an in-house team at roughly $458K to $593K a year, before you’ve found them. Hiring typically takes 6 to 7 weeks, plus about 3 months before a new DBA knows your systems.

For reference, our DBA as a Service starts at $11,950 per instance per year, with our Production SLA included.

A checklist before you sign

  • ☐ Response times are written down for each severity level
  • ☐ Coverage hours match when your customers use your product
  • ☐ Restores are tested on a schedule, with results shared with you
  • ☐ Minor and major upgrades are part of the service
  • ☐ Named engineers who know your environment
  • ☐ Support for community PostgreSQL wherever you run it
  • ☐ Onboarding starts with an assessment
  • ☐ Exclusions and extra charges are clear
  • ☐ You own your runbooks and documentation
  • ☐ The yearly cost is compared with the cost of hiring

Still weighing a service against a hire? Get a cost estimate in 2 minutes. It takes your instance count and coverage needs and shows both numbers side by side.

FAQ

What is a remote PostgreSQL DBA service? A team of PostgreSQL experts who manage your databases from outside your company under a service agreement. They handle monitoring, backups and restores, upgrades, tuning and incident response. It’s also called an outsourced DBA, managed DBA services or DBA as a Service.

Is a remote DBA service cheaper than a full-time DBA? For most teams, yes, especially when you need 24×7 coverage. One DBA covers about 40 of the 168 hours in a week; covering all of them takes at least three hires.

Can a remote DBA work with Amazon RDS or other managed PostgreSQL? Yes. Managed services handle the infrastructure, but query performance, schema design, upgrade planning and incident diagnosis are still yours. A DBA service covers that layer.

What should a PostgreSQL DBA service cost? It depends on the number of instances or cores, the coverage hours and the response times you need. Compare quotes as a yearly cost for your whole estate, not as a per-hour rate.

If you’re choosing a provider right now, I’d like to hear which of these questions got the most revealing answer. And if you’d like to see how our DBA as a Service answers them, talk to us.

Leave A Comment