Home/Methodology

How an analysis works with us

Transparent from the outset: clear requirements, a defined observation period, assessed results and an implementation plan that works in stages.

Initial conversation and scoping

We clarify the question, the scope (which machines, which partitions, which subsystems) and what data you already have. The result is a concrete proposal — not a standard package.

Outcome: scope, data requirement, schedule

Data provisioning

You provide operational data over a contiguous period — two full months as a guide, so that month end, weekends and batch cycles are included. Preparation runs without agents and without interfering with live operation.

The exact scope is agreed in the initial conversation and recorded in writing

Import and plausibility check

Data is imported, condensed and checked for completeness: gaps, interval changes, machine changes, anomalies. Anything odd is clarified before conclusions are drawn from it.

Journalling via checksums, no duplicate imports

Top-down analysis

From the machine via the partition down to service class, individual job and transaction. Two months as an overview, detail days at interval granularity. Peaks and smoothed curve always together — otherwise cost impact stays invisible.

Every analysis receives a professional assessment, not just a picture

Results workshop

A guided session in three parts: fundamentals so everyone speaks the same language; your data with assessment; and the recommendation and decision section. In German or English as you prefer.

Audience: systems technology, operations, procurement and management together

Staged implementation

Limits, weights and policy changes are adjusted step by step, each stage followed by measurement. That keeps it visible which change produced which effect — and keeps a rollback possible.

Optionally with recurring analysis to verify the effect

Principles

What we hold ourselves to

No finding without a data source

Every statement comes with what it was derived from. You should be able to reproduce every number yourself — even years later.

No recommendation without its price

Capping means accepting delay. Tightening goals means redistributing resources. We name the other side of every recommendation.

No number without provenance

We cite third-party studies with source and methodology. Where we cannot substantiate a figure, we do not use it — however good it would sound.

Your data stays your data

The analysis tool can run entirely on your premises. Where an extract is required, its scope is agreed in writing beforehand.

Prerequisites

What we need from you

Operational data over a contiguous period

Two full months as a guide. Which components are needed in detail depends on the question and is agreed beforehand.

The active WLM service definition

So that goals and classification are available in plain text rather than having to be guessed from measurements.

If required: a view into the subsystems

Transaction monitors, databases or Java environments — only where the question reaches that far.

For an audit: entitlement inventory and log data

Provided through the system's own utilities; the exact scope is agreed beforehand.

An analysis environment

Either on your premises — a simple server with PHP and a relational database is sufficient — or on ours, based on an agreed extract.

Does this fit your question?

In an initial conversation we clarify scope, data availability and schedule — with no obligation and no preparation needed on your side.