How network engineering services secure MACD changes
Article

How network engineering services secure MACD changes

A network engineering service secures MACD work (moves, adds, changes and disconnects) by running every request through the same six control points: validation at intake, a reviewed design, a pre-check with scoped access, execution inside an approved window, verification afterwards, and a record that matches the network. Each point leaves evidence an auditor can read.

Scroll Down

A network engineering service secures MACD work (moves, adds, changes and disconnects) by running every request through the same six control points: validation at intake, a reviewed design, a pre-check with scoped access, execution inside an approved window, verification afterwards, and a record that matches the network. Each point leaves evidence an auditor can read. The security comes from doing this for the small changes too, because routine MACD tickets are where controls get skipped, access gets left open and records drift away from the device.

Why MACD is where security slips

Large changes get attention. A core migration has a project manager, a change advisory board slot and a rollback plan that three people have read. A port move for a branch office gets a ticket.

That ticket is where the risk sits. Uptime Institute’s 2026 resiliency survey found networking and connectivity to be the most common cause of IT service outages, at 23%, and it ranks configuration and change management failures among the leading causes of network outages (reported by Network World, 14 May 2026). Its Annual Outage Analysis 2026 (13 May 2026) names failures to follow established procedures as the leading driver of outages caused by human error.

MACD volume makes that worse. A request queue of several hundred changes a week invites shortcuts: a shared admin login, a firewall rule opened for the change and never closed, a VLAN added on the device but not in the inventory. None of those causes an outage that day. Each one is a gap someone finds later.

The six control points

Step The control The evidence it leaves
1. Intake The request is validated against the inventory and the customer record before anyone designs anything. A mismatch stops the ticket. Validated request, inventory snapshot
2. Design A second engineer reviews the design and the configuration before the window, with the security policy checked for the affected segment. Peer-reviewed MOP, policy check
3. Pre-check Access is granted for this change only, on named accounts, and the device state is captured before the first command. Access approval, pre-change baseline
4. Execution The change runs inside the approved window, line by line from the MOP, with a named rollback owner on the bridge. Session log, window record
5. Verification Post-checks compare the device against the baseline and the design, and the service is tested end to end. Post-change diff, test result
6. Closure Temporary access is removed, the inventory and diagrams are updated, and the ticket closes only when the record matches the network. Access removal, updated records

NIST’s Guide to a Secure Enterprise Network Landscape (SP 800-215, November 2022) makes the case for treating the enterprise network as a set of controlled trust boundaries. In MACD work, steps 3 and 6 are where that becomes practical: access exists only for the life of the change.

What changes when a services team runs it

The customer’s process is the process. A services team that brings its own change model creates a second set of records, and a second set of records is a gap. The team works inside your change calendar, your templates and your go or no-go criteria.

Repeatable changes are automated. Port turn-ups, VLAN adds and standard firewall objects follow templates with pre- and post-checks built in, so volume does not depend on how tired the engineer is at 02:00.

Exceptions go to a senior engineer. The legacy device the script does not know, the order that changed after design and the customer with a non-standard handoff all need judgment.

Coverage follows the window. Most US change windows open overnight Eastern time, and a team that is working its day shift during those hours makes fewer mistakes than one working its night shift.

Five numbers to ask for

Ask whoever runs your MACD queue, in-house or outsourced, for these five figures, defined and dated.

  1. Failed change rate: changes rolled back or re-opened, divided by changes executed, per month.
  2. Rework rate: MACD tickets touched again after closure within 30 days.
  3. Interval: request received to change verified, median and 90th percentile, by change type.
  4. Access exceptions: temporary accounts or rules still open after the change closed, per quarter.
  5. Record accuracy: devices where inventory and running configuration disagree, from the last audit.

If a provider cannot give you the fifth number, it is not closing step 6.

Where Mikkena fits

Mikkena Global J.S.C. runs MACD, provisioning and migration work for US telecom operators and enterprises from Pristina, Kosovo, inside the customer’s change process. We publish a 99.99% provisioning SLA and price the work per change or per port with a rework guarantee, so the cost of a failed change lands on us. On one US Tier 1 carrier’s migration program, rework went from 28% to 0%. For a global FSI client we onboarded more than 900 workloads across 120 branches and three data centers, with zero trust segmentation. Across our data center and enterprise work, 80% of the infrastructure is managed as code. Our day shift covers 02:00 to 11:00 Eastern. Send us one month of MACD volume and we will price it per change within a day.

Sources: Uptime Institute, Annual Outage Analysis 2026 (13 May 2026); Network World on Uptime’s 2026 resiliency survey (14 May 2026); NIST SP 800-215 (November 2022).

Ask AI about this article

Pick a prompt. Your assistant gets the article and the context.

Why MACD is where security slips, the six control points and the five numbers to ask for.

Opens in a new tab with the prompt filled in.