Услуга · Custom software development

System integration

In words it all looks trivial: information should flow from one system into the next. The snag is that the reference lists do not match, and a machine cannot work out where the truth is recorded - people do that.

Precedence
each entity has its own
Failures
visible at once
Control
on how exchanges run
Manual entry
disappears

What the work includes

The technical part is nearly always solvable. The real work is agreeing inside the company which system counts as the master for each kind of data.

Discuss the scope

Survey

Which systems are involved, which data overlaps, and what is moved by hand today.

Scheme

What goes where, how often, and who is the master for each entity.

Mapping

We map product items, partner records and units of measure between the systems.

Implementation

Through the standard programming interfaces, or through a layer of our own if there are none.

Fault tolerance

What happens when a system is unreachable: retry, a queue, alerting a person.

Control

We watch it run and tell the responsible person when the exchange has stopped.

How it goes

A straightforward data transfer we set up in about two weeks, while linking several systems with shared reference lists takes six weeks and more.

01

Data

We study the reference lists on both sides. As a rule duplicates and discrepancies come to light that have to be cleared up before the start.

02

Project

The transfer scheme, precedence rules, the order for resolving conflicts.

03

Implementation

Development on a test environment with real data rather than invented data.

04

Launch

At first alongside the manual transfer, after which the manual one stops.

Data edited in two systems at once is the main source of conflicts. A sales manager corrects a client phone number on the site, an accountant does the same in the ledger, and the exchange overwrites one with the other. That is why an owner system is assigned in advance for every entity, while in the second one the field is locked against editing.

Questions and answers

Put things in order before the exchange, and that is a job for whoever keeps the records, not for a programmer. Joining up databases full of duplicates and discrepancies merely spreads the mess across both at once.

Very rarely. Most tasks are served well by once an hour or a few times a day, and that arrangement is noticeably more reliable. Instant exchange is needed where a decision is made right now, for example when checking stock at the moment an order is placed.

Workarounds come to the rescue: transfer through files, direct access to tables, replaying operator actions. Each is inferior to the standard route and calls for care during updates, yet the result is usually achieved.

We will join your programs up

Write which systems you move data between by hand. We will price the work and say what needs tidying up beforehand.