Data Warehousing — WhereScape RED — SQL Server

Your data.
One version.
Trusted.

Automated data warehouse builds using WhereScape RED — from initial design through to production-grade pipelines your team can operate and trust. SQL Server performance diagnosed at root cause, not patched.

Discuss Your Data Platform WhereScape RED
50× ETL runtime improvement in production
10+ Years — longest-running warehouse still live
6k+ Tables consolidated in a single estate
The Problem

Data trapped in
operational systems
costs real money.

Every month, senior people in your firm spend time assembling numbers that should already exist. Data pulled from NavOne, Business Central, and other systems — reconciled manually in Excel — before anyone can review actual performance.

That is not a reporting problem. It is a data architecture problem. The fix is a warehouse that pulls from your source systems automatically, reconciles at load time, and delivers numbers your Finance Director can sign off on without a weekend of preparation.

01
Month-end takes days instead of hours

Data from multiple systems consolidated manually. Same process, same risk of error, same cost in senior time — every single period.

02
No single version of the truth

Finance has one number. Operations has another. Compliance has a third. All defensible from their own sources. None of them reconciled.

03
Overnight jobs that nobody monitors

ETL runs complete — but silently drop records, skip partitions, or time out on large tables. Nobody knows until a report is wrong.

04
Regulatory data assembled by hand, every year

FATCA, CRS, and CARF require clean, auditable pipelines. Manual assembly carries compliance risk and escalating cost as reporting obligations expand.

05
Performance problems that survive every fix attempt

Slow queries, long-running jobs, I/O spikes at peak load. Patched repeatedly. Never traced to root cause. Still happening.

WhereScape RED

Warehouse automation
from design to production.

WhereScape RED automates the build of data warehouse objects — staging tables, dimension tables, fact tables, ETL procedures — from metadata. What takes weeks to hand-code takes days with RED.

But automation only delivers value when it is implemented correctly. Poorly configured RED environments produce automated bad code at scale. The difference is in the patterns, the Pebble templates, and the defensive ETL practices that separate a production-grade warehouse from one that works in development and fails under load.

Yitron has worked with WhereScape RED across the full lifecycle — greenfield builds, rescue projects, performance overhauls, and documentation of estates where the original author has left.

WhereScape RED lifecycle

01

Source Analysis

Profile source systems. Identify data quality issues before they enter the warehouse.

02

Model Design

Dimensional model designed for your reporting requirements — not generic best practice.

03

RED Build

Automated object generation. Custom Pebble templates eliminate manual code adjustments.

04

Test & Harden

Defensive patterns, error handling, job monitoring, selective data cloning for UAT.

05

Production

Monitored pipelines. Runbooks. Operable by your team without the original author.

Pebble Templates

Custom code generation at scale

Standard RED templates generate functional code. Custom Pebble templates generate code that matches your standards — error handling, logging, and naming conventions baked in at generation time, not added manually afterwards.

Defensive ETL

Pipelines that fail visibly

ETL that fails silently is worse than ETL that doesn't run. Every pipeline built by Yitron includes explicit error trapping, row count validation, and alerting that surfaces problems before reports are produced.

Job Monitoring

Know what ran, what didn't, and why

Custom job reporting layer on top of RED's scheduler — duration trends, failure history, and data volume tracking. Spots degradation before it becomes an incident.

Selective Cloning

UAT that reflects production

Custom utilities for selective data cloning — copy a representative subset of production data into UAT without exposing full client datasets. Test against realistic volumes without compliance risk.

Rescue & Recovery

Inherited RED estates

When the original author has left and the documentation doesn't exist — Yitron audits, documents, and stabilises existing RED environments. Then improves them. Starting with whatever is causing the most pain.

Handover

Work your team can maintain

Every engagement ends with documentation your team can actually use: data flow diagrams, object inventories, runbooks, and handover sessions. No dependency on Yitron to keep the lights on.

In Practice
50× ETL runtime improvement

A production performance problem that the full internal team had investigated and closed as unresolvable. Traced to root cause and fixed in a single engagement.

100× Query performance improvement

Fact table redesign prototype — benchmark improvements an order of magnitude beyond the existing structure.

10+ Years in continuous production

Data warehouse built for a Channel Islands organisation — still running without major rework.

Case Study — Channel Islands Financial Services

The ETL job that took 6 hours.
Then took 7 minutes.

An overnight ETL pipeline at a Channel Islands financial services firm had been running progressively slower for months. By the time Yitron was engaged, the job was consuming six hours of the overnight window — leaving no margin for reruns or delays before the business day started.

The internal team had investigated. Indexes had been rebuilt. Statistics updated. Query plans reviewed. The job continued to slow. It had been marked as a known issue with no identified fix.

6h Runtime before
7m Runtime after
30s Disk latency spikes
1 Configuration defect

The root cause was a SQL Server configuration defect causing 30-second disk I/O latency spikes on every write operation. Not a query problem. Not an index problem. A configuration problem — invisible to the standard diagnostic approaches the team had followed.

Single fix. 50× improvement. The pipeline has run within its window every night since.

More Case Studies →
SQL Server Performance

Root cause.
Not another patch.

Most SQL Server performance problems are diagnosed at the wrong level. A slow query gets a new index. The index helps for a week. The problem returns. The real cause — a configuration setting, an I/O bottleneck, a blocking chain, a statistics problem — is never found.

Yitron diagnoses SQL Server performance at every level: query execution plans, index design, statistics, tempdb contention, I/O subsystem behaviour, memory pressure, and server configuration. Including problems that are masked by successful job completion — where the job finishes but takes three times longer than it should.

Diagnose My Performance Problem
01

Establish the baseline

Capture actual runtime, I/O, CPU, and wait statistics under normal load before touching anything. Without a baseline, you cannot measure whether a fix worked.

02

Identify wait types

SQL Server records exactly what it is waiting for. PAGEIOLATCH, CXPACKET, LCK_M, WRITELOG — each points to a different class of problem. This is where most investigations skip to solutions too early.

03

Trace to configuration level

Query-level fixes address symptoms. Configuration-level fixes address causes. Max degree of parallelism, cost threshold for parallelism, memory settings, tempdb file count, instant file initialisation — these matter.

04

Validate the fix, not just the symptom

A fix that improves one query while degrading three others is not a fix. Every change is validated against the full workload baseline before it is recommended for production.

05

Document what was found and why it matters

Findings delivered as a written report — root cause, evidence, fix applied, and what to monitor going forward. Not just a list of changes made.

Further Reading

More from the
production floor.

The 50× ETL improvement is one example. The full case studies page covers the consolidation of 6,000+ tables across 20 databases, a data warehouse that has run for over a decade without major rework, and a KYC screening system that cut processing time by 20×.

All anonymised. All from production environments.

View All Case Studies →
Quick reference
50×
ETL runtime — configuration defect causing 30-second I/O spikes, missed by full internal team
100×
Query performance — fact table redesign prototype, benchmark improvement an order of magnitude beyond existing structure
6k+
Tables consolidated — 20-database estate, single consolidation layer for group financial reporting
10+
Years in production — Channel Islands data warehouse, continuous operation, no major rework
Next Step

If your data platform
is not working the way
it should —

A 30-minute call is usually enough to understand whether there is a structural problem worth fixing, or a performance issue worth diagnosing. No charge. No obligation beyond the call itself.

Book a Discovery Call

enquiry@yitron.co.uk +44 7829 800454 Jersey, Channel Islands