Summary

• Swisscom reports a 60% reduction in software upgrade time and a 97% reduction in network service setup time.
• Ericsson is introducing GitOps-based lifecycle and configuration management across its cloud-native dual-mode 5G Core functions.
• The deployment supports more automated operations, while wider AI-driven network autonomy remains under development.

Swiss telecommunications operator Swisscom has reduced network software upgrade time by 60% and service setup time by 97% through automated software management developed with Ericsson, demonstrating how established cloud engineering practices can change telecommunications operations.

The improvements come from introducing GitOps-based lifecycle and configuration management into Swisscom’s cloud-native network environment. Ericsson is making the approach available across its dual-mode 5G Core software, extending techniques used in cloud infrastructure to systems responsible for delivering mobile telecommunications services.

Swisscom has also changed how its engineers manage routine software releases, moving away from monthly maintenance windows and manual overnight upgrades towards automated deployment and configuration work during normal working hours.

The reported results relate to Swisscom’s implementation rather than a guaranteed improvement for every operator adopting Ericsson’s software. The company has not published a detailed breakdown of the original activity durations, the individual systems measured, or the sample sizes used to calculate the reductions.

Managing networks through approved configurations

GitOps is an approach to managing software and infrastructure in which an approved configuration is maintained in a version-controlled repository. Automated tools compare the operating environment with that recorded configuration and apply changes intended to keep the two aligned.

This changes the sequence of work involved in updating infrastructure. Rather than relying on engineers to make numerous individual adjustments, operators can prepare and review changes centrally before automated management tools implement the approved configuration.

The version history also creates a record of changes, supporting investigations into unexpected behaviour and allowing engineering teams to identify differences between the configuration that should be running and the one actually in operation.

Those differences, commonly called configuration drift, can develop when systems are changed outside an approved process or when deployments do not complete consistently. Detecting drift is particularly useful in environments containing many interconnected software services.

Telecommunications operators have been adopting cloud-native architectures that separate network functions into software components running on shared computing infrastructure. These arrangements can make services easier to update individually, but they also introduce additional configuration relationships that must be managed reliably.

Ericsson’s implementation introduces Git-compatible software packages and declarative configuration management for its cloud-native dual-mode 5G Core network functions. The capabilities are intended to work alongside its existing orchestration and in-service software upgrade systems rather than replace every established network management tool.

In practical terms, a software change can be reviewed, approved, and implemented through a repeatable process that records the intended outcome. Automated systems then help ensure the network maintains the required configuration after deployment.

Measurable reductions in operational effort

Swisscom’s reported improvements concern two activities that can require substantial coordination between engineering teams: upgrading network software and setting up network services.

Shorter upgrades may reduce the time engineers spend preparing, executing, and checking changes, while faster service setup can make it easier to introduce or modify network functions without repeating extensive manual procedures.

The reductions are reported as 60% less upgrade time and 97% less setup time. They should not be confused with a claim that overall network performance improved by those percentages or that customer-facing services became proportionally faster.

The original durations and operating conditions are not disclosed in the available technical material, limiting direct comparisons with other networks. A provider whose existing procedures are already highly automated may have less opportunity to achieve the same relative improvement.

Nevertheless, the Swisscom results show that software management practices can affect the operational work required to maintain a network. The operator’s move towards normal-hours deployment also indicates a change in how its engineering teams organise routine maintenance.

Philipp Bichsel, executive vice-president of Mobile Network & Services at Swisscom, has linked the deployment to the company’s wider adoption of cloud-native operations and continuous software delivery.

Operating changes still require suitable safeguards. A configuration error distributed automatically across several systems can produce wider consequences than an isolated manual mistake, particularly when the affected services support essential telecommunications functions.

Testing, access permissions, staged deployment, monitoring, and recovery procedures therefore remain necessary. A version-controlled configuration provides useful evidence of intended changes, but returning to an earlier software state may also require attention to application data and dependencies between services.

Automation creates a foundation for autonomy

Ericsson is incorporating the new management capabilities into its Intelligent Core offering, which is intended to support more autonomous network operations through software-defined objectives and AI-assisted analysis.

Automated configuration provides one of the mechanisms required for that approach. Monitoring systems can detect operating conditions and identify possible changes, while lifecycle management software provides a controlled route for applying an approved response.

The more difficult step is determining which changes a system should make without direct human intervention. Network management decisions can involve competing requirements for reliability, capacity, security, and performance, making the consequences of an incorrect action potentially substantial.

AI-supported recommendations may help engineers interpret operating information and select responses, but the appropriate level of autonomy depends on the task and the safeguards available.

Swisscom is exploring further applications, including AI-assisted automation of roaming partner configurations. Those activities are prospective developments and should not be presented as completed autonomous network deployments.

Roaming arrangements can involve multiple operators and technical agreements, requiring configuration changes that must remain consistent across organisations. Automating appropriate parts of that work could reduce repetitive administration, although the systems involved must still apply the correct contractual and technical requirements.

Deployment across established networks

Ericsson plans to extend its GitOps capabilities across its cloud-native core portfolio and connect them with its wider automation architecture. The approach is designed to support integration with products from different suppliers, an important consideration for telecommunications operators running networks assembled over many years.

Existing infrastructure may contain software and equipment with different management interfaces and upgrade procedures. Applying consistent configuration practices across those environments can require additional integration, particularly where older components cannot be managed through modern cloud interfaces.

Operators will also need to decide which deployment processes can be automated without changing the technical responsibilities associated with essential services. Maintaining a clear record of authorised changes provides a foundation for that work, but governance remains an operational responsibility.

Swisscom’s implementation has produced measurable reductions in two network management activities, while Ericsson is broadening availability of the supporting software. Further deployments will establish how consistently the approach performs in networks with different architectures and operational procedures, and whether it supports the more autonomous functions now being developed.