Database Professional Services
Yourdatabasestopsbeingthethingnobodywantstotouch.
A database operations team embedded in your workflow — sizing and tuning the workload, proving failover instead of hoping, testing recovery on a schedule, and reporting attainment quarterly against the tier you contracted.
- PostgreSQL
- Oracle
- SQL Server
- MongoDB
- Redis
- TimescaleDB
- PostgreSQL
- Oracle
- SQL Server
- MongoDB
- Redis
- TimescaleDB
Who this is for
The situation that makes this urgent
If none of these describe you, this practice is probably not your next spend — and we would rather you found that out here.
A production database whose performance nobody can explain, on an engine one person understands.
A recovery objective written in a policy that has never been tested end to end.
Regulated month-end, disbursement or settlement cycles where a slow query becomes a reportable event.
Three ways to engage
Assess, Transform, Operate
The same three shapes on every practice page, so a client who has read one page can navigate another instantly.
Assess
A fixed-price baseline of where the estate actually is. Yours whether or not you proceed.
Transform
Remediation and re-platforming against the vendor-supported pattern for your engine.
Operate
Ongoing operations at a contracted tier, with a named engineering group rather than a queue.
Capability detail
The technical substance, on the page and not behind a form
Named engines, versions, and the patterns we operate them under. This is where the engineer decides.
PostgreSQL
Streaming replication with Patroni; quorum-based failover
- Versions operated
- 13 – 17
- Replication lag
- < 5 s
Oracle
Data Guard, maximum availability mode; RAC where licensed
- Versions operated
- 19c, 23ai
- Replication lag
- < 2 s
SQL Server
Always On availability groups, synchronous within region
- Versions operated
- 2016 – 2022
- Replication lag
- < 3 s
MySQL / MariaDB
Group replication or semi-sync with orchestrated failover
- Versions operated
- 8.x / 10.6+
- Replication lag
- < 5 s
MongoDB
Replica set with majority write concern
- Versions operated
- 6, 7
- Replication lag
- < 2 s
Redis
Sentinel or cluster mode, persistence policy set per workload
- Versions operated
- 7
- Replication lag
- n/a — cache tier
TimescaleDB
Native PostgreSQL streaming replication; continuous aggregates and compression policies kept in sync across replicas
- Versions operated
- 8.x / 10.6+
- Replication lag
- < 5 s
Snowflake
Multi-cluster warehouses with automatic scaling; cross-region replication and failover groups for disaster recovery
- Versions operated
- 6, 7
- Replication lag
- < 2 s
How we work
Four phases. Objective exit criteria.
A gate passed on assertion has not been passed. One exit criterion per gate is shown here; the full set sits on each practice page.
Discover
Workload profile, dependency map and recovery objectives written down and signed off by the client owner.Design
Target architecture reviewed against the vendor-supported pattern for the engine in question; failure modes enumerated.Deliver
Recovery tested from backup within the stated RTO, evidenced by a timestamped test record.Operate
Runbooks in place, rota published, first quarterly service review held with measured figures.
The boundary
What we do not do
Stated because a capability list without a limit reads as marketing.
Out of scope
Application-layer schema ownership and business logic changes. We advise on them; we do not own them.
Partner-delivered
Vendor licence resale and licence position audit defence, are handled with a commercial licence negotiation remains with the client.
Declined
Unsupported end-of-life engine versions on production, unless the engagement includes a dated upgrade path.
Evidence
Reference architecture, in place of a case we cannot yet name
No client has consented to publication yet. Rather than an unattributed metric, here is the artefact.
Two-region PostgreSQL topology, financial services workload
Synchronous standby in-region, asynchronous replica cross-region, quorum failover, and the specific failure modes each element does and does not cover.
Recovery test record — redacted sample
The form the client receives after each scheduled restore: objective, measured result, deviation and the action taken.
Measurement window and method stated on every published figure.
Get started
A two-week discovery produces your database baseline: profile, HA gaps, an executed recovery test and a costed coverage model.
You keep all four documents whether or not we work together.