Can you secure a contact center's IT without stopping it?

Yes, one piece at a time: first measure the real state (workstations, accounts, access), then run the new alongside the old and switch off the old only once the new is proven.

Why open source hosted on-site instead of cloud?

It's autonomy as well as cost: the data stays in the company and no external vendor can switch off a service or change the terms.

Where do you start with a tangled infrastructure?

With a measured census: how many workstations are alive, how many accounts are real, who accesses and how. Every decision backed by a measurement.

Taking back a contact center's IT, one piece at a time

A contact center working for several utility companies (water, gas, energy, waste) lives on three things: the phone ringing, the ticketing system and the operators' workstations. When one jams, everything stops. Our client had reached the point of no longer knowing how their own infrastructure was built. Here is how we put it back in order, without ever stopping work.

01First rule: measure, don't assume

Faced with a tangled infrastructure, the temptation is to start over. We did the opposite: before touching anything, we measured it. Of 41 registered workstations, 22 were alive; the rest were leftovers from a larger organization. In the ticketing system, 36 active accounts out of 137: the others belonged to people gone for years. A probe recorded access every fifteen minutes. The rule: every statement backed by a measurement.

Intuition tells you where to look, not what to do. More than once a reasonable assumption proved false against the facts.

02The things nobody could see

The census surfaced problems no one had ever looked for: the ticketing system had had no working backup for years (millions of tickets left to chance); on every workstation three remote-access programs installed and forgotten; the system passwords in cleartext, dozens of accounts sharing one; years of attachments gone from disk. Not glaring mistakes: the cumulative effect of years with no overall picture.

03The strategy: unpick, don't tear

Much of the infrastructure was run by outside vendors. Instead of opening disputes, we took back control piece by piece, documenting every step:

a real corporate VPN on the office firewall instead of improvised access; backups, with an on-site server receiving a nightly copy of the system; centrally managed workstations (they install the company software themselves, configure the browser, block unauthorized extensions); remote support moved to an open-source tool hosted in-house that logs who connects; the internal chat, also open source and on-site; a rack inventory. Each piece introduced alongside the old, tested, and only then the old switched off.

Additive before destructive. Open source hosted in-house is a choice of autonomy: the data stays in the company and no vendor can switch a service off.

04Tools, not heroes

What we're proudest of: every repeatable operation became a tool. Checking system health, installing on every workstation, verifying who's on VPN, resetting a workstation: written, tested procedures that don't depend on one person's memory. That's the difference between infrastructure that works because someone is good and infrastructure that works because it's built well.

05The results

Within weeks: the system has verified daily backups on-site; half the useless workstations and accounts removed, shrinking the exposed surface; all remote access goes through tracked corporate tools; the company has living documentation of its infrastructure. And above all: work never stopped.

06What we learned

1. Measure first, then decide. 2. Unpicking beats tearing: piece by piece, no leaps in the dark. 3. Open source in-house is autonomy, not just cost. 4. One written tool is worth ten brilliant interventions.