Skip to content
DBS

Database solutions

Design, migration, performance and recovery — mostly SQL Server, and mostly for systems whose data has outgrown its original design.

  • SQL Server
  • Database migration
  • Performance optimisation
  • Backup and recovery
  • Data protection
  • Microsoft Azure
Code
DBS
Class
C · Under 6 months
Engagement
TYPICAL 1-4 MONTHS · AUDIT FIRST
Stages
04
Deliverables
06
Sections
06

Overview

Applications get rewritten; data outlives them. A schema designed for a company a tenth of the current size is the most common performance problem we are called about, and it is rarely fixed by a bigger server. We design new schemas, repair old ones without downtime, tune the queries that actually matter, and make sure the recovery story is real. Ravensbourne Property’s reporting database went from a nightly job that overran into the working day to a fifty-minute run, with no hardware change — the fix was three indexes, one schema correction and removing a view that had been calling itself.

Benefits

05 points
  • Tuning driven by measurement, not intuition. We work from actual wait statistics and query plans, and we tell you which changes will not help.

  • Migrations executed online where the platform allows it, with a rollback that has been tested rather than described.

  • A recovery objective per database and a restore drill that proves it, timed and dated.

  • Schema changes shipped through the same pipeline as code — versioned, reviewable and repeatable — instead of applied by hand at midnight.

  • Archival and retention designed deliberately, so tables do not grow without limit and audit obligations are met by policy rather than by nobody deleting anything.

Workflow

04 stages
  1. Health assessment

    One to two weeks. Wait statistics, the worst queries by total cost rather than by duration, index usage and duplication, growth trend, and the current backup and restore position. The output is a ranked list with expected impact.

  2. Quick wins

    The changes that are safe, reversible and measurable go first — usually indexing, statistics and a small number of query rewrites. This funds the rest of the work politically as well as financially.

  3. Structural work

    Schema corrections, partitioning, archival, or a platform migration. Executed incrementally with the application running, and validated with row counts and checksums rather than a glance.

  4. Recovery and monitoring

    Backup policy against a stated objective, an executed restore drill, and monitoring that alerts on growth, blocking and failed jobs before a user notices.

Deliverables

06 items
  • Assessment report with ranked findings, measured impact and the changes we advise against.
  • Versioned migration scripts, applied through your pipeline, with rollback scripts alongside.
  • Index and query change log, with before-and-after measurements for each.
  • Backup, retention and archival policy, plus dated evidence of a completed restore.
  • Monitoring and alerting configuration for growth, blocking, deadlocks and job failure.
  • Data dictionary and entity model, kept in the repository next to the schema.

Questions

03 entries
  • Sometimes, but check the cheaper answers first. A large share of slow reporting is a handful of missing indexes, statistics that have not updated, or reporting queries competing with transactional load — that last one is often solved by a read replica rather than a warehouse. We measure before recommending, and a warehouse is a real commitment of both money and people.

  • For most SQL Server migrations, yes: replicate, verify with row counts and checksums, then cut over in a short maintenance window with a tested rollback. True zero downtime is achievable and costs meaningfully more; we will price both and let you decide which the business actually needs.

  • If you are on SQL Server and it works, stay. It is an excellent database and you already have the licences and the skills. We recommend PostgreSQL for new systems where licensing cost matters and there is no Microsoft-stack dependency, and a document or cloud database only when the access pattern genuinely calls for it rather than because it is newer.

Book a consultation

01 locations

Complete IT, software and AI solutions

  • Indore, India