Unlocking IoT & AI Capabilities with Edge Data Centers

ByWang Bo

November 29, 2019

Wang Bo

The commercial use of 5G will bring massive IoT and AI applications that will change the way we work and live, but it will also require much more in terms of network capabilities. 

The essence of IoT lies in the intelligent network and the core of IoT is to connect things and manifest their intelligence. However, we can only do so by combining big data and IoT. As everything becomes connected, computing and cloud capabilities will be enabled to accelerate the commercial use of AI.

What Is the Basis of Ubiquitous Connectivity?

To answer this question, let’s look at the standard definition of multi-access edge computing (MEC):

“A network architecture concept that enables cloud computing capabilities and an IT service environment at the edge of the cellular network.”

Edge computing is a new phase of industry digitalization that drives the development of cloud computing. In essence, it extends cloud computing to the user edge and creates a new form of cloud computing.

With the emergence of various new service types, such as AR/VR and autonomous vehicles, the traditional network structure is gradually becoming overburdened. Multi-access Edge Computing (MEC) is emerging as tool that enables network services to “sink” to the radio access network side of users.

Service “sink” has the following advantages:

1. Decreases the transmission latency perceived by users.

2. Controls network congestion.

It works by providing hardware infrastructure for IT+CT+OT (such as computing, storage, network, and intelligence) and software cloud services (such as IaaS, PaaS, DaaS, and SaaS) near data sources or users.

A traditional data center is basically a centralized big data processing platform, while MEC is an edge-based big data processing platform.

Under this architecture, a traditional data center is divided into various small data centers that are placed at the edge of a network. Data centers can be closer to users, providing faster services and achieving better network performance.

Obviously, ubiquitous connectivity relies on edge computing. According to statistics, about 70% of IoT data is processed at the edge.

Edge computing with such a large space and large capacity will need to meet the construction requirements of hardware infrastructure for IT+CT+OT, such as computing, storage, network, and intelligence, to be near data sources or users.

So, where should we deploy the basic hardware infrastructure for IT+CT+OT (such as computing, storage, network, and intelligence)?

In edge DCs.

With massive IoT and AI applications, the number of edge DCs will also surge. However, they have particular requirements for construction and management:

1. Simple and flexible deployment

Simple installation and maintenance is required for a large number of edge DCs. Unlike large-scale DCs, edge DCs can use dedicated hardware to facilitate quick installation and flexible access. And they can carry MEC capabilities. In 2B and 2C scenarios, edge DCs of different sizes and capabilities need to be flexibly deployed to meet requirements in different scenarios, such as enterprise campus and street cabinets, for a plug and play experience.

2. Efficient operations and maintenance (O&M)

Since the number of edge DCs is high, the pressure and costs involved in O&M are also high. Therefore, unified O&M is essential to improve efficiency.

3. High reliability

High-speed computing features like the low latency of 5G and cloud computing require edge DCs to be highly reliable, for example, through prefabricating the core components of edge DCs in the factory and designing redundant components.

The Answer Is Modular

Based on these three requirements, modular DCs are the best choice for constructing edge DCs for ubiquitous connectivity. The modular DC is fast to deploy, can be flexibly expanded, and it matches the intelligent management system for unified O&M. It meets the requirements of edge DCs for simplicity, efficiency, and reliability.

As a leader in modular DC solutions, Huawei Network Energy has launched the modular DC solution FusionModule800 3.0 for edge DCs, building on the FusionModule800 2.0 integrated data center.

Modular design, fast deployment, and flexible expansion

Huawei FusionModule800 3.0 DC adopts a modular design where the core components are prefabricated and pre-commissioned before delivery. Complex onsite installation is left at the factory, so delivery is fast, onsite installation and maintenance are simple, and capacity can be flexibly expanded by adding modules, meaning that initial investment does not necessarily have to be exorbitant.

Smart O&M for efficiency gains

The data center infrastructure management system DCIM-NetEco6000 helps unify the management of massive volumes of edge DCs, improving O&M efficiency.

Huawei DCIM can implement dynamic visualized management for multiple DCs on one graphical user interface (GUI), helping users quickly identify the running status of DC infrastructure. The geographic information system (GIS) map function ensures that faults are located quickly, reducing data center risks. The mobile O&M app is provided to check DC running status anytime and anywhere. Even non-O&M experts can efficiently and accurately perform preventive maintenance inspections based on the operations guide.

To control access, FusionModule800 3.0 uses facial recognition technology, so that authorized users can easily log in to the local surveillance system and open any cabinet door through the local surveillance screen.

Cold and hot aisle containment reduces power consumption and improving efficiency

FusionModule800 3.0 uses cold and hot aisle containment to ensure that equipment is cooled before the ambient environment. It prevents dust and reduces noise, improving cooling efficiency. FusionModule800 3.0 also uses direct current inverter air conditioners, electronically commutated (EC) fans, and wet film humidifiers to improve cooling system efficiency and slash annual electricity fees.

Prefabricating core components in the factory and designing redundant components to ensure high reliability

The core components are prefabricated, preinstalled, and pre-commissioned in the factory, which reduces the potential risks caused by onsite cabling and commissioning. Strict factory inspections and testing certification also ensure the product reliability.

Leading the trend of DC modularization is only one of things we’re doing to evolve networks. In the future, considering the massive construction requirements and pain points of edge DCs, we will dive deeper into how the AI+Cloud platform can continuously create intelligent modular kernels, making DCs simpler, more efficient, and more reliable.

Read more about Huawei’s modular data centers.

Leave a Reply

Posted in Cloud Posted in Cloud
Published by

Wang Bo

View all posts >

Subscribe

Events

%d bloggers like this: