Understand, stabilise and improve complex IT legacy systems

01 Stabilise · Immediate Relief & Deep Tuning David vs. Goliath

SQL Server Performance & High Load (VLDB)

David versus Goliath: Small optimised server done in 1.2 ms versus burning monster server without optimisation
Tuned: 1.2 ms (8 Reads) Without: 17.8 s (3.46M)

Resolve slow screens, blocked nightly runs and deadlocks during live operation – by precisely measuring execution plans, indexes and locks instead of buying expensive hardware.

64 bn rows · 4 TB (NXP) ~130 Procedures refactored
For IT Management & Operations To “David vs. Goliath” ↓
02 Improve · Hands-on in the Engineering Team Hands-on in Team

Hands-On Team Coaching towards DevOps

BEFORE: Manual & Ad-hoc × prod_deploy.sql ! Lock Timeout ! No Rollback ! “Who changed Prod?” No Tests · Island Know-how WITH COACHING: Practical DevOps Git & PR Clean Code Docker CI Auto-Build NUnit & DB Checksums ✓ LIVE PRTG ✓ 142 NUnit & SQL Tests: PASS ✓ Checksum Verify: 100% OK Coach codes hands-on in your repo & team
Before: Ad-hoc & Risk After: CI/CD & Auto-Tests

No theoretical slide presentations: I roll up my sleeves in the code and pipeline – Azure DevOps CI/CD, Docker, automated NUnit/checksum tests and safe releases.

Hands-on in Code vs. slides CI/CD · Docker · NUnit · PRTG
For Engineering, CEOs & HR To “DevOps Coaching” ↓
03 Understand · Toolset COMPIDDA Extra Assessment & Toolset

X-Ray & Document Orphaned IT Landscapes

COMPIDDA Dependency Diagram: Complete X-ray of a grown software and database landscape down to the line of code
COMPIDDA · 100% Local Code → Procedure → Table

Before rebuilds, migrations or retirements: using my locally running toolset COMPIDDA, I deterministically analyse and document grown database and software landscapes down to the line of code (separately priced).

Deterministic (0 hallucination) Separate On-Site Assessment
Against island know-how & blind rebuilds To “COMPIDDA Analysis” ↓
Manuel Hagen · Diplom-Wirtschaftsinformatiker (FH) · 30+ years experience

Is your SQL Server too slow – or is a grown IT landscape slowing down your business? I resolve the bottleneck and make your system and team future-proof.

Screens and reports that take seconds. Deadlocks and timeouts. Nightly runs that are not finished by morning. Or critical legacy systems that nobody dares to touch any more. I resolve database bottlenecks measurably during live operation, guide your team hands-on in the code towards practical DevOps and X-ray your entire IT landscape with my proprietary toolset COMPIDDA in a separate on-site assessment.

Book a 20-min initial call → COMPIDDA Assessment & DevOps View lab measurements

Directly for IT leaders, executives, department heads & HR: You describe your situation (even from a business perspective without a technical brief). I tell you honestly where the leverage lies and what a sensible starting point looks like.

  • 64 bn rows · 4 TBProduction database & DevOps pipeline I was responsible for as app owner (NXP)
  • around 130 proceduresoverhauled during live operation, with a written client reference
  • ×358 fasterthanks to a matching index, measured in my own lab (10M rows)
  • 30+ yearsDatabases, controlling mindset, full-stack (.NET) & team coaching
Manuel Hagen

I am Manuel Hagen, a freelance business informatics graduate (Diplom-Wirtschaftsinformatiker FH) from Buxtehude near Hamburg. My background combines budget controlling (PHOENIX-AG), 14 years of FDA-regulated clinical study reporting (Clario/ERT) and responsibility for 64 billion records at NXP Semiconductors. You work directly with me – clear for management and business departments, deep in the code for your IT.

Drawing: on the left a small, quiet server among plants and a relaxed man with a coffee – the search is already done. On the right a huge, smoking, burning server and a desperate man wearing ear protection – still searching.
David versus Goliath · Why optimisation comes before hardware

Scale down. And still be faster.

Optimisation means: first measure where the time really goes – then fix it right there, instead of blindly buying expensive hardware and core licences.

  • Bought a new server – and it is still slow?
  • The overnight run is not finished in the morning – and everyone waits?
  • Nobody dares to touch the old programs any more, because nobody knows what depends on them?
  • What if your old server were faster than the new one?
  • Everything hangs in day-to-day business – and the test shows nothing?
  • More servers, more licence costs – and the wait stays?
  • Your server searches the whole haystack for the needle. With an index it reaches straight in.
  • How much more hardware will you buy before someone measures where the time goes?
  • Burning electricity on work nobody needs?
  • Small machine with an index: a blink. Big one without: 18 seconds.

Measured in our own lab

An example from the lab: weaker machine, still faster

The same search in a table of 100 million rows. Waiting time in seconds, shorter is better.

Strong machine without index8 instead of 2 processor cores, 10x the memory
17.8 s
Weak machine without indexBaseline
30.9 s
Weak machine WITH indexthe same weak machine
under 0.01 s
  • 4x the processor cores and 10x the memory made the search only 1.7 times faster – see measurement details ↓
  • With an index, the weak machine was 14,864 times faster than the strong one without – lab value, see measurement details ↓
  • Pages read: 3,465,122 without index, 8 with index.

Here it was a single missing index – one cause among several. Others: overnight runs that do not finish, blockages in day-to-day business, grown programs nobody dares to touch any more. See typical situations →

Green IT: Without optimisation the same search took about 22 seconds of computing time, with optimisation practically none. Less computing work means less load on the hardware (power consumption not measured).

If the problem lies elsewhere, or everything is fine, I will tell you. Better no assignment than the wrong one.

Request a performance check → See the exact measurements

How we measured

Where the 1.7 comes from: 30,934 ms (weak machine without index) divided by 17,837 ms (strong machine without index), each the mean of 3 runs with the cache cleared. Both machines were emulated on the same computer: the strong one used all 8 cores and 13,000 MB of memory, the weak one 2 cores and 1,300 MB, with only one core per query. Disk and memory bandwidth were the same for both. If a real, larger server also has faster disks, the gain from hardware may be larger – we did not measure that.

Where the 14,864 comes from: 17,837 ms (strong machine without index, mean of 3 runs) divided by 1.2 ms (weak machine with index, mean of 20 runs); the database cache was cleared before every run. Times around one millisecond are coarse – at 2 ms it would still be about 8,900 times. Depending on the comparison, the factor in this run ranges from about 3,900 (strong machine without vs. with index, 4.6 ms) to about 25,800 (weak machine without vs. with index). The reliable figure is the number of pages read: 3,465,122 vs. 8.

Lab with synthetic test data, cold cache. The machines were emulated on an 8-core computer; disk and memory-bus speed were not emulated. For Green IT, only processing time was measured: about 22 seconds of CPU time per search without optimisation, practically none with it. We did not measure power consumption. All measurement logs

Typical triggers

Do you recognise any of these symptoms?

Whether you look at it from the perspective of executive management, a business department or IT leadership: every technical symptom has a direct operational and commercial impact.

Response times

“Screens and reports take seconds to minutes.”

Technical approach:

Measure the execution plan and reads; fix missing or badly designed indexes, data type conflicts (e.g. nvarchar vs. varchar) and parameter sniffing. In the lab, a table scan dropped from 349,716 to 11 reads.

Business impact (Departments & Management): Smooth day-to-day work in order processing, warehouse and customer service without rewriting the application and without buying new servers.
Index measurement stage by stage →
Nightly run

“The nightly run is not finished by morning and slows down everything around it.”

Technical approach:

Break up oversized transactions, replace cursor loops with set-based SQL, eliminate lock chains. In client projects I have used this approach to cut nightly runs from several hours to just over an hour.

Business impact (Departments & Management): Reliable figures, financial closings and shipping releases ready when the morning shift starts at 7 a.m.
View four case studies →
Deadlocks & timeouts

“Blocking in day-to-day business that can never be reproduced in testing.”

Technical approach:

Analyse lock chains and dynamically assembled SQL, convert it into fixed, parameterised procedures with a stable execution plan, make transactions smaller, choose a suitable isolation level.

Business impact (Departments & Management): No aborted bookings, no halted production lines under peak load, and immediate relief for your internal IT support.
Case study: errors only in production →
Island knowledge & legacy code

“Nobody dares to touch the grown procedures and systems any more.”

Technical approach:

Make all dependencies visible with COMPIDDA, test recorded cases against the legacy code using checksums and disentangle step by step. Reference: around 130 procedures at an international spirits producer, during live operation.

Business impact (Departments & Management): Protection against knowledge loss when key staff retire or leave; new business requirements become plannable again.
Remediation approach →
Data growth & releases

“Data volume has exploded – or every team release turns into a nail-biter.”

Technical approach:

VLDB architecture (partitioning, insert-only, Redis queues) combined with hands-on DevOps team coaching: automated CI/CD pipelines, NUnit/checksum tests and clean deployments. Experience from 64 bn rows at NXP.

Business impact (Departments & Management): Stable 24/7 operation despite growth and a confident in-house team that delivers changes reliably without weekend outages.
Semiconductor production case study →
Hardware reflex

“The next suggestion is: a bigger server.”

Technical approach:

Measure first, then invest. In the lab (100M rows), a server with 8 instead of 2 cores and ten times the memory was only 1.7 times as fast without an index; with a matching index, even the small server was 14,864 times as fast as the large one without (lab value with synthetic data; times around 1 ms are coarse – at 2 ms it would still be about 8,900 times). I will also tell you when hardware is the right answer after all.

Business impact (Departments & Management): Avoids unnecessary five-figure spending on additional SQL Server Enterprise core licences that do not solve the root cause.
Calculation: index vs. licence costs →
Services & getting started

In three steps from suspicion to a measured solution

You decide after each step whether to continue. Before I implement anything, we put scope, effort and acceptance criteria in writing.

Step 1

Initial call

20 minutes · no obligation · phone or video

  • You describe your situation – whether an acute SQL bottleneck, a grown black-box landscape or desired team coaching towards DevOps.
  • I give you an honest initial assessment of the cause, leverage and commercially sensible approach.
  • No access to your systems needed; equally suitable for executives, business heads or HR without a technical specification.
Book an initial call →
Step 2 · Entry offer

Performance check

The proof on your own system

  • Measure where the time is lost: wait times, most expensive queries, locks.
  • Narrow down the cause in your most critical query or nightly run: execution plan, reads.
  • Rank measures by effect, each with a measurable success criterion and a way back; implement the most important directly applicable measure after your approval.
  • Clear scope boundary: If the root cause requires deeper code or architectural changes beyond the check’s scope, you receive the exact empirical proof, prioritisation and effort estimate as a sound basis for decision-making.
  • You receive: a summary for executive and IT management, technical findings with before-and-after measurements, a list of measures and all scripts used.
Request a performance check →
Step 3 · Deepening & expansion

Remediation, DevOps Coaching & COMPIDDA

As a standalone task, in your team or as an extra assessment

  • Index and query tuning across the board as well as remediation of grown stored procedures (with checksum tests).
  • Hands-on team coaching towards DevOps: CI/CD pipelines, Docker, automated tests and knowledge transfer directly on your real project.
  • COMPIDDA Assessment (separately priced): On-site assessment and deployment of my toolset to thoroughly analyse and document orphaned IT landscapes.
  • Architecture of very large databases (VLDB) and application modernisation (Oracle/Desktop → MS SQL & Blazor).
View all modules →
Review a concrete deliverable in advance

What do you get in writing after a Performance Check?

No vague slide decks: take a look at my complete Sample Report (PDF, in German) featuring an executive summary for decision-makers, before-and-after measurements, root-cause analysis and a concrete action list including rollback steps.

Open sample report (PDF) ↗
Measured instead of claimed

What targeted optimisation achieves: two lab measurements and one case study

Measurement logs from client systems stay with the client. That is why I reproduce typical causes in my own SQL lab: synthetic test data, no customer data, measured by script and repeatable at any time.

×358 faster

Missing index

10M rows: A covering index replaces the table scan with targeted index access – without new hardware and without changes to the application.

Before (scan)
349,716 reads
With index
11 reads (×31,792)

Stage overview · Benchmark data & execution plans →

3 instead of 27,469 reads

Wrong data type in the parameter

If the .NET application sends Unicode (nvarchar) to a varchar column with SQL collation, SQL Server has to convert every index entry and reads the entire index:

nvarchar (scan)
27,469 reads · 446 ms
Matching (seek)
3 reads · 2.7 ms

Ten further measurement series · Benchmark data on collation →

Practice · billion-row scale

High load in production

At NXP Semiconductors, around 1.4 bn new serial numbers per month were checked for duplicates against 64 bn existing rows – decoupled via Redis queues:

Direct (locks)
Blocking / timeouts
Redis queue
< 15 ms response time

Semiconductor production case study →

Case studies

From practice: starting point, approach, result

Six projects that show how I work across database performance, team coaching, IT landscape analysis and modernisation; further cases are on the Database Optimisation page.

VLDB · 64 bn rows · 4 TB

NXP Semiconductors: duplicate checking in semiconductor manufacturing

Starting point: As technical app owner, I was responsible for the quality assurance platform TDLog: 64 billion existing records (4 TB), around 1.4 billion new records per month from the production lines worldwide, which had to be checked for duplicates against the entire existing data in real time. Any delay would have held up production.

Approach & result: Insert-only architecture with temporal data storage, decoupled Redis queues for processing in under 15 ms, distributed satellite databases at Asian production sites with replication to headquarters. The 64 billion records were migrated in batches, in parallel with ongoing 24/7 production and without production downtime.

  • MS SQL Server
  • 64 bn rows · 4 TB
  • 1.4 bn rows/month
  • Insert-only & temporal
  • Redis queues
  • Docker microservices
  • Azure DevOps CI/CD
Remediation · around 130 procedures during live operation

International spirits producer: legacy MSSQL procedures overhauled

Starting point: Around 130 MSSQL procedures that had grown over the years for nightly data consolidation and error correction. Deeply nested logic and unclear dependencies led to long run times and blocked downstream systems.

Approach & result: The procedure landscape was systematically analysed, disentangled and largely replaced by maintainable, considerably faster T-SQL solutions – during live operation, without interrupting operations, safeguarded by checksum tests.

“Mr Hagen has exceptionally profound technical expertise in the areas of databases, application architecture and performance optimisation. […] His work significantly improved the transparency, maintainability, stability and performance of the IT systems in use.”

— from the client’s written reference letter (translated from German)

Approach to procedure remediation →

COMPIDDA · Dependency analysis Extra assessment & toolset

X-raying orphaned IT landscapes & resolving single points of knowledge

When key people leave the company or grown databases and applications are due to be rebuilt, often nobody knows exactly which procedures, views, jobs, SSIS packages and C#/VB applications read or write which tables. For this I offer a dedicated, separately priced on-site assessment with my COMPIDDA toolset: it parses up to 19 source types and visualises all dependencies with the exact source location.

Dependency diagram of a grown software and database landscape, created with COMPIDDA
Dependency diagram · drill-down to the source location (Read / Write / Execute) Full view ↗

The analysis runs entirely on your premises and works deterministically, without AI or cloud uploads. Anything that cannot be resolved unambiguously stays visibly marked; the business assessment remains with the people responsible on your side. Details on the COMPIDDA assessment →

Hands-on · Team Coaching & DevOps

Leading development & DB teams practically towards DevOps

Starting point: In grown IT teams, manual deployments, missing regression tests and historical fear of breaking production often block progress. Pure slide-deck consulting does not change day-to-day work.

Approach & result: As a calm, pragmatic senior engineer directly inside the team (“a steady rock in rough seas”), I meet developers and administrators where they are. Together we build automated Azure DevOps CI/CD pipelines, containerised test environments (Docker), reproducible NUnit and SHA-256 checksum tests for database changes, and meaningful monitoring (PRTG) – including structured onboarding of new colleagues.

  • Azure DevOps CI/CD
  • NUnit & checksum tests
  • Docker
  • PRTG monitoring
  • Team coaching & onboarding
14 years · Regulated environment (FDA)

Clario / ERT: Clinical study reporting & validation

Starting point: Over 14 years of freelance work for US medical technology and pharmaceutical service providers (ERT / Clario). Clinical respiratory study data from international sites had to be consolidated reliably, reproducibly and auditable for regulatory submissions (including FDA requirements).

Approach & result: Complex study reports, ETL pipelines and validation logic developed and maintained in MS SQL Server and Oracle PL/SQL – in English-speaking, globally distributed project teams with strict traceability and schedule discipline.

  • Medical / Pharma
  • FDA-regulated
  • MS SQL & Oracle PL/SQL
  • SSRS / Study reporting
  • International teams (EN)
Oracle → MS SQL · Blazor

Import wholesaler: desktop application with Oracle brought to the web

Starting point: A VB/WinForms application with an Oracle database that had grown over more than 20 years was to be replaced. The goal was a Blazor web application with MS SQL Server.

Approach & result: Existing components and use cases analysed independently, new architecture planned and implemented, the new functions verified with NUnit and Selenium in a dedicated test environment. The business workflows were preserved; the technology was replaced.

  • Oracle → MS SQL
  • C# / Blazor
  • Entity Framework Core
  • NUnit / Selenium
Process

Measure, narrow down, agree, fix, re-measure

Every engagement follows the same sequence. At any time you can see what I am doing, why, and how it can be rolled back.

  1. 1

    Measure

    Where is the time being lost: CPU time, input/output, locks or network? The basis is system views, query statistics and execution plans – not the contents of your tables.

  2. 2

    Narrow down

    Usually a few queries, procedures or jobs cause most of the load. I examine these individually: execution plan, reads, wait times, locking behaviour.

  3. 3

    Agree

    You receive the measure with its expected effect, risk and way back. Nothing is implemented until you approve it – gladly together with your team or your service provider.

  4. 4

    Fix

    Preferred diagnostic strategy: index before code before hardware. We safeguard larger changes in advance with automated tests and checksums over recorded cases. For every step there is a state you can switch back to.

  5. 5

    Re-measure & hand over

    The same measurement as before, by script. Your team receives the result, the reasoning and the documentation so that it can carry on without me.

Security & ground rules

What you should know in advance as the person responsible for IT

Your data stays with you

The diagnosis only reads system views, query statistics and execution plans (VIEW SERVER STATE permission), no table contents. You can review the diagnostic scripts in advance. I work only on your systems; I do not accept data exports, backups or test data by email. I do not pass on customer data, schemas or measurement logs.

No intervention without approval

I only implement changes to indexes, procedures or configuration after your consent, in line with your change process and with a prepared way back.

Access on your terms

Remotely via the access you provide, via screen sharing in which you keep control, or on site by arrangement. On request, I will sign a non-disclosure agreement (NDA) before my first look at the system.

Directly with the expert

No sales team, no intermediary layer, no changing contact person: you have the initial call with me, and I am also the one who does the work.

Knowledge stays in-house

I explain causes and measures in a way that lets your team understand them and recognise them itself in future. Scripts, tests and documentation are part of the result.

Honest assessment

If the problem is not in the database, or a standard tool is enough, I will say so. Better no assignment than the wrong one.

About

Manuel Hagen – database expert, DevOps coach & business informatics graduate with an eye for ROI

Background & commercial mindset

I am Manuel Hagen, a freelance database expert, full-stack engineer and business informatics graduate (Diplom-Wirtschaftsinformatiker FH). For more than 30 years I have been bringing together databases, software, team processes and business efficiency. You work directly with me – from the first measurement to the handover to your team.

Why executive management and business departments also understand me: My career took me from budget controlling at PHOENIX-AG into database and application development; my part-time degree in business informatics at FOM Hamburg deepened this dual perspective. For every technical measure, I calculate what it costs, what operational risk it reduces and what licence or hardware investment it saves.

At NXP Semiconductors I worked as a Software Engineer and FullStack Software Engineer. As technical app owner for the global quality assurance platform TDLog (64 billion rows), I was responsible for database architecture, C#/.NET backend, automated tests and Azure DevOps deployment – and coached the international team in architecture, planning, PRTG monitoring and onboarding.

As a freelancer, my work has included 14 years of clinical study reporting in FDA-regulated environments (ERT / Clario), overhauling grown procedure landscapes during live operation (Stockspirits), modernising desktop applications with Oracle to MS SQL and Blazor, and developing my deterministic analysis toolset COMPIDDA to X-ray and document orphaned IT landscapes.

  • 01 Measure first, then change.

    No measure without measuring before and after – verifiable in numbers.

  • 02 Commercial clarity.

    Preferred diagnostic strategy: index before code before hardware – translates technical metrics into business impact for management.

  • 03 Respect what exists.

    Grown systems are made visible with COMPIDDA and improved step by step, not replaced on reflex.

  • 04 Hands-on team coaching.

    No slide consulting: I help directly in the code and pipeline until your team works safely on its own.

At a glance for executives, procurement & HR

Key engagement facts for decision-makers & project staffing

Request CV / project profile →
Commercial benefit (CEO / CFO)

Avoids five- to six-figure SQL Server licence and hardware upgrades, eliminates production downtimes and secures knowledge before key staff retire.

Contract & compliance (Procurement)

Direct contract (freelancer, no agency margin required) or via your preferred provider. NDA in advance, diagnosis without table contents, clear scope packages.

Qualification & industries (HR / Recruiting)

Diplom-Wirtschaftsinformatiker (FH), 30+ years IT. Proven in 24/7 semiconductor production (NXP), FDA pharma/medtech (Clario/ERT), trade/logistics & industry.

Core stack & roles

Senior SQL Performance Architect, DB/DevOps Coach, Legacy Moderniser. MS SQL (T-SQL, SSIS, SSRS), Oracle PL/SQL, C#/.NET, Blazor, EF Core, Docker, Azure DevOps.

References

What clients and employers have put in writing

Permanent positions at NXP Semiconductors and PHOENIX-AG, freelance projects in semiconductor production, medical study reporting and business applications.

Getting to grips with complex systems

“… able to quickly get to grips with even complex, historically grown system landscapes and to improve them sustainably. His work significantly improved the transparency, maintainability, stability and performance of the IT systems in use.”

Excerpt from the reference · international spirits producer

Clear for both IT and business

“Particularly noteworthy is his ability to consistently combine technical modernisation with organisational processes and to communicate IT topics in a way that is both technically sound and understandable to the business side.”

Excerpt from the reference · international spirits producer

Reliable in 24/7 operation

“Mr Hagen was reliable to a very high degree … but above all, time and again, he showed full commitment when it came to ensuring operations.”

Excerpt from the employment reference · NXP Semiconductors

On time and clear in coordination

“He completed his work on time and he has a good communication with his lead on the status of each deliverable he is assigned with.”

Excerpt from the reference · Clario (USA)

  • NXP SemiconductorsFormer employer · TDLog
  • ClarioMedical · formerly ERT
  • ERTMedical · study reporting
  • VeotecMedical
  • Stockspiritsformerly Borco
  • PHOENIX-AGFormer employer · Controlling & IT
  • Cremation servicesCrematorium process control

Short excerpts from written references; the German originals are translated into English, the Clario excerpt is quoted in its original wording. Omissions are marked with “…” or “[…]”. Full documents are available on request.

Beyond pure SQL tuning

IT landscape analysis with COMPIDDA, hands-on DevOps coaching & modernisation

Fast queries are the start. When historically grown architectures, missing documentation or manual team workflows slow your organisation down, I support you with two specialised focal areas and solid software modernisation.

Specialised service · Toolset deployment Extra assessment · separately priced

COMPIDDA Assessment: X-raying & documenting orphaned IT landscapes

When documentation is missing, key developers are about to retire (“island knowledge”) or a legacy replacement is planned, gut feeling is dangerous. With my proprietary software toolset COMPIDDA, we systematically map your grown database and application landscape on your premises.

  • Up to 19 source types read automatically: MS SQL Server, Oracle, PostgreSQL, SSIS packages, SSRS reports, SQL Agent jobs, C#/.NET, VB, PowerShell, configuration files and more.
  • Typed dependencies with exact source location: Every connection is classified as reading (R), writing (W) or executing (X) – drill-down right to file and line number.
  • Bottom-up architecture view: Even thousands of objects remain readable through clean aggregation by servers, databases, schemas and applications.
  • 100% local & deterministic: Runs on your premises without cloud uploads and without AI hallucinations. Ambiguous dynamic accesses are flagged honestly for review.
Step 1 · Technical scan (e.g. Day 1 on site) Set up read-only connectors, parse schemas, procedures, jobs and code repositories locally into the COMPIDDA dependency graph.
Step 2 · Impact analysis & documentation handover (e.g. Day 2) Joint review of critical bottlenecks, dead objects and blast radius with your team; handover of the interactive documentation.

Note on pricing: A full IT landscape analysis with the COMPIDDA toolset goes beyond a standard SQL performance check and is commissioned and priced separately as a dedicated customer assessment including toolset deployment.

Request a COMPIDDA assessment
Team enablement · Hands-on in code Practical instead of theoretical

Team coaching towards DevOps: A steady rock in rough seas

Many development and database teams are stuck in day-to-day firefighting: changes are deployed manually, refactoring stored procedures feels risky, and new colleagues take months to become productive. I do not coach from the sidelines with slide decks – I work alongside your team in the codebase.

  • Pick up the team where it stands: Respectful collaboration with existing DBAs and developers without finger-pointing or dogmatic buzzwords.
  • Automated safety nets for database & code: Building NUnit integration tests and deterministic SHA-256 checksum tests over real data snapshots before refactoring legacy procedures.
  • CI/CD & reproducible environments: Setting up clean Azure DevOps pipelines, version-controlled database schemas and Docker-based test environments.
  • Active monitoring & onboarding: Establishing early-warning metrics (e.g. PRTG sensors for wait stats, deadlocks and job runtimes) and structured onboarding for new team members.
During the engagement Critical performance and architecture roadblocks are resolved together directly on your production-relevant tasks.
After the handover Your team deploys faster, safer and independently – with automated regression tests and clear runbooks.
Discuss team coaching

Modernising applications

Moving desktop and Oracle applications step by step to MS SQL Server and Blazor without losing business workflows. Implemented with C#/.NET and Entity Framework Core, safeguarded by automated NUnit and Selenium tests.

Data integration & migration

Interfaces, ETL pipelines (SSIS, REST, CSV/XML/JSON, SFTP) and VLDB migrations carried out in stages during live 24/7 operation with a verified way back.

Procedure & architecture remediation

Disentangling hundreds of grown T-SQL or PL/SQL procedures into maintainable, fast modules – safeguarded by automated before/after checksum verification.

Tools & stack: MS SQL Server (T-SQL, SSIS, SSRS) · Oracle PL/SQL · PostgreSQL · MySQL · Redis · COMPIDDA · C#/.NET · Blazor · Entity Framework Core · NUnit · Selenium · PowerShell · Docker · Azure DevOps · PRTG. The choice follows the problem; what works well is integrated rather than replaced.

Frequently asked questions

What IT managers, executives & project leads ask beforehand

Do you need access to our production data?

Not for the diagnosis. System views, query statistics and execution plans show where time is lost without me reading table contents. For changes such as new indexes, we follow your approval process; if you prefer, your team implements the change itself and I accompany it.

What does a performance check cost – and how is the COMPIDDA assessment priced?

The 20-minute initial call is free of obligation. For the SQL Performance Check, we agree scope and a fixed package price in writing in advance. If you want to X-ray and document an entire grown or orphaned IT landscape across databases, jobs and application code using my COMPIDDA toolset, this is a separate on-site assessment including toolset deployment that is priced separately (typically starting with a compact 2-day assessment). For comparison: a bigger server with additional SQL Server core licences quickly costs many times as much – see the calculation on the Database Optimisation page.

How does hands-on team coaching towards DevOps work in practice?

I do not hold abstract methodology lectures. Instead, we pick a concrete pain point in your database or application project – for instance fragile releases, untestable stored procedures or missing monitoring. Together with your developers and DBAs, I build the automated tests (NUnit, SHA-256 checksums), the Azure DevOps CI/CD pipeline or the Docker test environment directly on your real codebase until the team feels confident running it every day.

Do you work remotely or on site?

Both, by arrangement. Performance diagnoses and ongoing project work can usually be carried out via a secure remote connection that you provide; COMPIDDA landscape assessments, architecture workshops and team coaching handovers preferably take place on site. I am based in Buxtehude near Hamburg.

And what if the problem is not the database at all?

Then I will tell you openly. The measurement shows whether the time is lost in the database, the network, the C#/.NET application or the hardware. Because I cover both database architecture and full-stack .NET development, we can solve application-side bottlenecks (such as Entity Framework Core patterns) right away as well.

What if the application has to keep running 24/7 during the work?

That is the normal case for me – from semiconductor manufacturing to spirits logistics. Changes are made in small steps, each with automated tests in advance and a state you can switch back to. Critical areas come last, once the approach has proven itself.

Contact & project start

Choose your starting point – 20-minute initial call without obligation

Whether you are an IT manager with an acute SQL bottleneck, a managing director facing a legacy black box, or an HR/project lead looking for a seasoned senior expert: select the matching focus below or write to me directly.

Path 1 · Acute bottleneck

SQL Performance Check

Queries, reports, batch runs or deadlocks are slowing down operations? We isolate the root cause by measurement and rank the top fixes.

Select Performance Check
Path 2 · Extra assessment & toolset

COMPIDDA Landscape Assessment

X-ray and document an orphaned or historically grown IT & database landscape on site (up to 19 source types, separately priced assessment).

Request COMPIDDA Assessment
Path 3 · Enablement & rebuild

Team Coaching, DevOps & Modernisation

Hands-on coaching towards CI/CD & automated testing, procedure remediation or Oracle/.NET application modernisation.

Select Coaching / Project
Enquiry

Prepare the initial call

What is it about? multiple selection possible

100% privacy-friendly: Your entries are only checked locally in your browser and are not transmitted to any web server. Clicking opens your email client with the pre-filled text – or simply write directly to manuel.hagen@datrexx.de.

Initial call