The growing prevalence of digital transformation across industries and the emergence of innovative new technologies such as 5G and IoT have seen service cloudification gradually mature. This poses higher requirements on network quality, such as bandwidth capacity, reliability, security, and openness. On top of this, conventional deployment and O&M methods based on engineer experience and command lines are facing great challenges. Against this backdrop, more and more enterprise IT teams have found that, to some extent, such human-based network construction and O&M mode prevents us from fully unleashing the performance of network devices. This presents the biggest obstacle to eliminating network bottlenecks. As such, there is an urgency to achieve network intelligence that enables networks to support on-demand change and automatic recovery. The underlying difference between the datacom network and other IT infrastructure is that new technologies must be innovated based on legacy technologies. That's because any network technology upgrade must be implemented at the lowest risk without affecting services. Every IT team is exploring how to evaluate risks and new technologies to smoothly upgrade traditional networks to intelligent ones. For this, my thoughts are as follows:
First, set actual service requirements as the basic principle and ultimate goal of network evolution. This means that we need to analyze the network based on service requirements and pain points, classify various requirements by technology update, process optimization, and personnel requirements, and specify the direction and objectives of network intelligence. Take campus networks as an example, office networks are mainly deployed in traditional campuses, and on such networks, email, conference collaboration, and FTP are typical services. With the development of IoT technologies, a series of smart campus service requirements are posed. To meet this, campus networks should not only carry office services. For example, production lines of a manufacturing campus need to connect to various detection devices, warehousing AGVs, air conditioner temperature sensors, and other devices. In this case, the campus network carries the functions of the production network. In public areas like government service halls and airports, the campus network of enterprises and institutions are available to the public for free. In this case, the campus network serves as a service network. These examples indicate that service changes determine the functions of campus networks and the direction of network intelligence. When there are production requirements, to achieve network intelligence, further improvements on ubiquitous connectivity, reliability, and deterministic quality of the network should be considered. Moreover, strengthened cooperation between production-related departments and IT teams throughout the process is worthy of our consideration. When service requirements exist, network construction should focus on network openness, security, as well as rights- and domain-based management. These technical indicators are key to the decision-making in terms of network technologies. And not only that, network O&M issues, such as complaints about user experience, long application onboarding time, and hard-to-find fault locating tools, should be also considered during target network construction. The resolution of such issues needs be set as one of the network construction objectives.
Second, evaluate network reconstruction risks and determine the technology upgrade path based on the direction and objectives. When shifting service objectives into technical requirements, we need to consider feasibility. To achieve that, we can start from classifying the risk of changes in descending order. For high-risk changes, such as those related to architecture (SDN or not), physical connection mode, network parameter change (such as IP and VLAN), and network protocol (such as IGP, BGP, and MPLS), we need to preferentially consider building a new network before migrating the legacy network. For low-risk changes, such as those related to security hardening and O&M tool enhancement, direct upgrades can be performed on the legacy network. However, the interface support status and performance compliance of legacy network devices must be considered. Some new technologies, such as UCL (security group policy) and Telemetry (performance collection), require hardware support. As for transforming office campuses towards production and service campuses, we can first determine the SDN-based network architecture, preferentially use wireless technologies such as Wi-Fi 6/7 and RFID as the main access mode, and introduce switches and O&M tool platform that support various intelligent features. Then, we can run the newly-built network and the legacy network concurrently before gradual service migration, for all-round network optimization.
Third, develop network health checklist, determine the baseline, and consolidate network intelligence results in real time. As services are continuously being updated, network performance gradually deteriorates. For a wired and wireless converged network, the IT team must create a health table for network indicators from dimensions, to evaluate and monitor network quality. These network indicators consist of the access success rate, access duration, coverage and interference, roaming fulfillment rate, capacity fulfillment rate, and throughput fulfillment rate. Then, the IT team can obtain the normal baseline based on the tool platform and specify the events and thresholds that require manual intervention. When network indicators deteriorate, the platform is preferentially used for self-optimization to shorten response time. To some extent, this shifts passive handling into proactive prevention.
Intelligence will drive all networks. Only by embracing changes, analyzing gaps, and taking the right steps to utilize new technologies throughout processes can we build networks that are adaptable and purposeful.
Disclaimer: The views and opinions expressed in this article are those of the author and do not necessarily reflect the official policy, position, products, and technologies of Huawei Technologies Co., Ltd. If you need to learn more about the products and technologies of Huawei Technologies Co., Ltd., please visit our website at e.huawei.com or contact us.
