PostgreSQL care and support
Managed PostgreSQL. Without hiring a DBA.
We look after production databases for organisations where a database outage stops the work. Monitoring, backups proven by actual restores, tuning, migrations and a support contract with a guaranteed response time. We work in every industry, and we know healthcare deepest, having run hospital databases for twenty years. A database that survives an emergency department will survive your peak season.
We start with a thirty minute call, free of charge. We learn your environment and tell you plainly whether we are the right people for the job. Only after that does anything cost money.
Sentences we hear most often
We hear them in hospitals, in software houses and on factory floors, because a database does not ask about your industry. If any sounds like your situation, a conversation makes sense. None means someone was negligent. This is what a database looks like when nobody owns it full time.
"Backups run every night, but nobody has ever checked whether we can restore from them."
"The database slows down roughly once a month and we do not know why. A restart helps."
"We are on an old version, support is ending, and we are afraid to migrate in production."
"The person who knew our database has left. They left one page of notes."
"The auditor asks about log retention and encryption, and we have nothing to answer with."
"An in-house DBA costs roughly PLN 15,000 a month fully loaded. For one person, who also takes holidays."
What we cover
Seven areas. You do not have to take all of them at once, but we always do the first two, because without them the rest is built on sand.
Monitoring and incident response
Round the clock machine supervision of availability, replication, disk space and long running queries. Alerts reach an on-call engineer, not a mailbox nobody reads at night.
Backups and verified restores
Backups are half the work. The other half is restoring them regularly on a separate environment and measuring how long it took. Only then do you know your real RTO and RPO.
Tuning and query optimisation
We look for the cause of a slowdown instead of adding more memory. Query plans, indexes, table bloat, autovacuum settings. We measure the effect before and after, on your queries.
Version upgrades and migrations
Leaving an unsupported version, moving from Microsoft SQL Server or Oracle to PostgreSQL, relocating between data centres. Rehearsed on a copy, with a rollback plan, before we touch production.
Replication and high availability
Replicas, automatic failover, connection pooling. We build as much as your acceptable downtime requires, not as much as fits into a proposal.
Database security review
Who has access and from where, what is encrypted, what ends up in logs and for how long, and whether the database answers the entire internet. The last one happens more often than companies expect.
Ad hoc support
Consulting hours without a standing contract. A review before a rollout, a second pair of eyes on a schema design, help with an incident already under way.
Three service levels
The times below are the same values we put into our service contracts. We do not promise more on a website than we sign in an annex.
| Level | What it includes | Response time | Fix or workaround |
|---|---|---|---|
| Supervision | monitoring, alerts, monthly report | 4 hours during service hours | agreed per ticket |
| Care | plus security patches, tuning, planned maintenance window | 15 minutes during service hours | critical 4 hours, blocking 1 business day |
| Critical | plus critical failures around the clock, recovery plan and its tests | 15 minutes during service hours, 2 hours outside them | 4 hours during service hours, 6 hours outside them |
Service hours: business days and Saturdays, 07:00-21:00 Central European Time. Outside that window we handle critical failures through an emergency number provided in the contract. For critical and blocking failures we report progress every 30 minutes, then hourly. We roll back to the previous version within one hour of your request. Every month you receive a report with ticket statistics, times and any breaches.
A support contract for open source
PostgreSQL has no vendor you can write a letter to. That is its strength, right up until an auditor, a regulator or the other side's legal team asks the question: who is accountable for this database.
We sign a contract that answers it. We do not sell a licence, because PostgreSQL is free and stays free. We sell accountability for it.
A named accountable party
The contract names a company with a registration number, deadlines and penalties for missing them. Not a forum, not a mailing list, not "the community". There is someone to call and someone answerable.
An agreed version and its lifecycle
We agree which version we maintain and for how long. You do not have to migrate every year just because a new one shipped. The date you leave a version is written down, not decided by whoever notices first.
Security fixes within contracted windows
We track vulnerabilities in PostgreSQL and in the extensions you actually run, and patch within the window agreed with you. You hear about a vulnerability before your auditor finds it.
A document for the auditor
A support contract closes the question of vendor support for a system processing sensitive data. Under GDPR, NIS2 and national cybersecurity rules that question always comes, and "we use open source" is not an answer.
A report you can show
Every month you receive tickets, response times, patches applied and backup status. That is evidence of due diligence, not an internal note of ours.
No vendor lock-in
The database stays plain, open PostgreSQL. We do not install proprietary extensions it would stop working without. When you want to leave, you take the data and the database whole, and the contract ends with its notice period.
How working together looks
About three weeks pass between the first contact and full care. The times given are when you receive something, not how many hours we spend.
Call
We learn the environment, the data volume and what hurts most. We say whether we are the right fit at all.
30 minutes, freeAudit
Remote review of configuration, backups, performance and access. No access to personal data required.
PLN 2,500 net, credited against your first invoice data collected within 1 business dayReport
Risks ranked by weight, with a recommended order of fixes. The report is yours, with no strings attached.
within 5 business days of accessRemediation
Monitoring, restore-tested backups, urgent fixes from the report. You choose the scope.
up to 10 business daysOngoing care
Continuous supervision at the chosen level. First monthly report after thirty days.
from the end of step 4We enter environments that hold your data
It is sharpest with patient data, and that is where we learned to work, but we apply the same rules everywhere. We say them plainly rather than in a footnote. Below are only things we have, not things we plan.
Data processing agreement
Signed before the first access to any environment. We do not start work on a promise that the document will follow later.
Information security policy aligned with ISO standards
NIS2 and national cybersecurity requirements are daily practice in our hospital work, not a topic to learn on your project.
Accountable engineer access
Every entry into your environment is logged and sessions are recorded. It is clear who did what and when, including when you must show it to an auditor.
Professional liability insurance
PLN 500,000. The question about insurance comes up in every serious procurement, so the answer is here.
Encryption at rest and in transit
This covers backups as well, and the channels we use to reach you.
Findings we did not go looking for
If we notice something that threatens your security while working, you get it in writing, even when it falls outside our scope.
What we have already done
No wall of logos here. Two pieces of work that describe the scale and the kind of problems clients bring us.
Migration from Microsoft SQL Server to PostgreSQL
- Situation
- Databases of the Polish National Health Fund running on the Microsoft engine, with rising licence costs and rising load.
- What we did
- Moved them to PostgreSQL, together with reworking whatever stopped behaving the same way after the migration.
- Outcome
- Hundreds of thousands of zloty saved on licences and maintenance services, with better stability and a faster environment.
Database infrastructure for a medical cloud service
- Situation
- A cloud medical service where every customer has a separate database, plus several dozen synchronised databases at OpenCare software customers.
- What we did
- Turned managing that infrastructure into a process: provisioning, updates, backups and supervision performed identically for every instance.
- Outcome
- Running several dozen databases at once stopped depending on whether somebody remembered a particular server.
Who does the work
PostgresCare is run by the OneCare team, the same people who build and maintain medical systems for Polish healthcare providers. We look after a database the same way whatever runs on top of it.
Over twenty years with PostgreSQL
Since migrations off version 9, and throughout on systems where downtime means a queue of patients, not an unavailable shop.
The engine known from the inside
Not from a course and not from a cloud provider console. From two decades of fixing other people's decisions and our own, in production.
We know hospitals from within
We know how a maintenance window in a clinic differs from one in a hospital with an emergency department, and why migrating on a Friday is a bad idea. That schooling travels well.
We are not selling you a system here
We look after the database you already have, whatever software runs on it and whoever supplied it.
Let us start with a conversation
Thirty minutes, free, no slide deck. Tell us what is broken and what worries you, and we will tell you whether this is a job for us.
Do you have a service contract with us?
Emergency tickets go through the channel named in your contract, and the emergency number for reports outside service hours is in its text. We do not publish it here, because a phone number visible to everyone stops being an emergency line.