As data pervades all aspects of our lives, we are moving into an era that can only be described as software defined everything (SDE), where the value within data networks shifts decisively from hardware to software. For the users, it means that the hardware—server, storage, or network—is no longer provisioned manually by IT staff.
Listed below are the key technology trends impacting the software defined everything (SDE) theme, as identified by GlobalData.
5G deployments are capital intensive compared to earlier mobile generations. The use of often higher frequency spectrum bands and expected increases in both data volumes and connected devices require significant network densification through small cells.
Software defined networking (SDN) and network function virtualisation (NFV) bring cost savings and improved operational efficiency for operators. These technologies are key elements of 5G implementation and are already being deployed by many operators.
Traditionally, data centre hardware was specialised and operated on closed, proprietary standards.
White box devices are generic, non-branded hardware and have increasingly replaced proprietary hardware in the data centre. They are developed by smaller manufacturers and are cheaper than their proprietary counterparts. They may come pre-loaded with open-source software or may be sold as a bare metal device.
Merchant silicon or off-the-shelf chips replace custom silicon in data centres. Custom silicon, usually ASICs (application-specific integrated circuits), is designed by the company selling the switches in which they are used.
Traditional networking infrastructure was controlled by proprietary software running on proprietary hardware. SDN decouples the software layer from the underlying hardware and allows the use of open standards. This has resulted in the rise of open source network controller software. Unlike proprietary software, anyone can modify and enhance open source software (OSS).
OpenFlow was the pioneer in open source SDN and was adopted by many telecom operators and cloud providers. This was followed by the launch of the Open Networking Lab (ON.Lab), which released the Open Network Operating System (ONOS) source code (SDN controller) to the open source community.
In principle, software defined architecture provides a means to automate virtual security measures for the data centre.
The security functions, such as firewalling and intrusion detection, are decoupled from the proprietary hardware. This allows the data centre administrators to update and manage security from a central point using software tools.
Where compliance with local regulations is a factor, appropriate security protocols can be applied specifically to the appropriate sensitive data as required.
Wide area networks (WAN) are rarely cloud-friendly. Many branch offices often do not directly connect to a cloud service but are routed to headquarters first. This leads to latencies that impair the performance of in-branch cloud-based applications.
SD-WAN is based on a centralised software control function that securely and intelligently directs WAN traffic. It optimises the connections between geographically distributed locations to reduce latencies and improve software performance.
Intent-based networks (IBNs)
IBNs are SDNs where the software can make real-time decisions and implement changes to the network. The concept is that humans tell a network an intent and the network software figures out what action is needed to fulfil that intent.
The IBN relies on machine learning (ML) and artificial intelligence (AI) to fulfil the intents. Once the intent has been defined, IBNs, guided by AI frameworks, continuously self-monitor and automatically fix problems or achieve real-time results.
SDN architecture automates network configuration and management, but it is still based on static code. It takes the predefined hard-coded commands to take action. SDN benefits from the use of AI and ML, which provides the software with learning and decision-making abilities. The AI-based SDN effectively solves the problem on the machine end and reduces the need for human intervention.
At the 2019 Mobile World Congress (MWC), Telefonica showed its framework for intelligent connectivity.
Hyperscale internet data centres
Hyperscale data centres run more than 100,000 servers and are linked through high-speed fibre-optic networks. The cloud providers and internet giants, Amazon, Microsoft, Google, and Facebook, have been bringing hyperscale data centres on stream over the last few years.
As software is detached from hardware in an software defined data centre (SDDC), it allows these companies to remove expensive proprietary data centre hardware providers like HPE, Dell, and Cisco.
SDNs have become an integral part of cloud-based IT (information technology) infrastructure. As enterprises increasingly embrace cloud services, SDN provides a centralised way to manage data and workload between distributed locations.
The new SDN offerings are not only for private data centres but are integrated with public clouds. For example, the SDN controller of Cisco’s Application Centric Infrastructure (ACI) uses native APIs offered by public cloud providers and integrates with them.
Silicon photonics is an emerging technology that combines laser and silicon technology on the same chip. It allows data to be transferred between computer chips by optical rays, which can carry far more data over longer distances in less time than the electrical conductors traditionally. This supports faster interconnects between data centres.
Photonic chip technology is still in the research phase. The costs of manufacturing silicon integrated circuits with embedded photonic elements have held back progress in the technology.
Intel has emerged as a leader and has demoed its 400G silicon photonics transceiver at its Interconnect Day event in 2019. At the same time, Cisco exhibited its silicon photonics platform at the Optical Fiber Communication Conference.
This is an edited extract from the Software Defined Everything – Thematic Research report produced by GlobalData Thematic Research.