Modern enterprises need high-performance, scalable and efficient data centre fabrics, says Houman Modarres, Head of Enterprise Marketing at Nokia. 

Data centres are the critical infrastructure that drives modern digital enterprises. But organisations in every vertical are assessing their data centres in new ways as they step up their efforts to innovate and ensure that the infrastructure meets the needs of cloud and legacy applications that are the lifeblood of their businesses. To get the maximum value from these applications while handling the massive volumes of data they generate, enterprises need data centre solutions that can elevate the performance, scalability and efficiency of their operations.

Businesses might worry that striving for rapid digital and cloud innovation will overtax their data centres and put their mission- and business-critical operations at risk. At some level, these concerns might be well founded, but action is needed.

Data centre performance that’s adequate for today’s enterprise applications won’t be good enough to support AI/ML, VR cloud, IoT, AI, 5G or industrial metaverse applications. It’s time for new enterprise data centre networking approaches that can adapt to the demands of any application and drive innovation with greater openness, extensibility and automation.

Adapting to new data centre networking challenges

Rapidly evolving business needs are driving IT operations to roll out new applications faster. Consequently, IT operations teams have embraced data centre architectures leveraging Kubernetes, containers, microservices and DevOps processes using CI/CD methods for high agility, reliability, and hardware efficiency. A growing enterprise also needs data centres capable of accommodating increased capacity, storage, users, and geographical reach. These concurrent realities place tremendous pressure on the data centre networking infrastructure.

Cloud engineering and IT teams have responded to this need by deploying modern leaf-spine architectures, or data centre fabrics. These architectures provide the scalability and bandwidth enterprises need to meet growing demands within the data centre and to interconnect to their clouds.

These innovations and architectures have helped companies take the next meaningful step in their cloud transformation journey. But these operations and cloud teams still struggle to balance the need to develop new applications and services that will help grow the business with the need to perform the routine operations, troubleshooting, maintenance and administrative tasks required to keep the business running.

Data centre solution vendors have added automation capabilities to their platforms to help operations and cloud teams achieve this balance. Recognising that companies want to move away from proprietary closed systems and speed up innovation, some vendors now offer partially open network operating systems (NOSs), along with limited automation and third-party application support. These are steps in the right direction, but enterprises need more capabilities and flexibility to adapt to new demands with speed and confidence.

Building next-gen data centre fabrics

Enterprises are seeing the opportunity to gain a competitive advantage by building intuitive, elastic data centres that can deliver greater scale and value and handle new types of workloads with more agility. But they can’t make this happen with closed or partially open NOS solutions that are difficult to operate.

Businesses need truly open, next-generation data centre networking solutions that will make it easy to design, deploy, adapt, operate and automate data centre network fabrics at scale.

The essential foundation of a next-generation data centre fabric is a completely open Linux-based NOS that is designed for agile, flexible operations. An ideal NOS will enable greater scalability, performance, and efficiency across all data centre network operations by empowering data centre networking teams and application developers with:

• A cloud-native design approach centred on flexibility, programmability and resilient IP routing to make network operations more agile and adaptable
• An open telemetry framework that can meet the granularity and scalability demands of any application
• A publish/subscribe architecture that uses open-source innovations to share state information
• Hardened protocol stacks that deliver robust IP routing performance, interoperability and security, including Ethernet VPN (EVPN), Multiprotocol-Border Gateway Protocol  (MP-BGP) and Virtual Extensible LAN (VXLAN), along with a model-driven management architecture that can maximise visibility and simplify operations and third party integrations
• A microservices-based, state-efficient design that bolsters hitless enhancements to applications enhancements and delivers always-on networking
• A programmable, open-source command line interface (CLI)
• A standard, unmodified Linux kernel that provides a foundation for building a suite of network applications.
  

A next-generation data centre fabric NOS must also provide automation tools that will help clear a route to fully automated operations. For example, declarative, intent-based automation tools can enable network teams by programming and implementing network automation tasks at scale. By doing so, it can drastically reduce the time and effort spent designing, deploying and operating data centre fabrics. Tools that use open frameworks and microservices can help enterprises apply network operations (NetOps) approaches to boost efficiency at every phase of the data centre fabric lifecycle.

Enterprises will face new challenges as they introduce new data centre automations, applications, integrations, and workloads. To help mitigate risks, next-generation data centre fabric solutions should provide a safe emulation environment such as a digital twin of the data centre network. This will allow NetOps teams to test changes, see their effects and make adjustments before implementing them in the live network. In this manner, they can move super-fast but with full confidence.

High-performance hardware is a must for any enterprise seeking to get maximum value from a fully open NOS and advanced automation toolkit. Modern data centres need switching platforms that can deliver port speeds up to 400 Gb/s, massive scale, superior reliability and seamless interconnectivity. These platforms should support fixed and modular configurations for top-of-rack, leaf, spine and super-spine applications.

The applications should also support a wide range of capabilities across Layers 2 and 3. By having platforms that combine merchant silicon with a common NOS and hardware design across multiple form factors, the modern NOS provides flexibility that enables data centre and cloud networking teams to meet the demands of any application.

Getting the edge

Enterprises need next-generation data centre fabrics to take on the challenges and opportunities presented by cloud and hyper-digital services. A solution that combines a fully open NOS with intent-based automation tools and high-performance hardware will address this need by enabling businesses to quickly and easily scale their data centre fabrics to address changing business requirements; take control of data centre fabric operations by capitalising on new levels of openness; and automate at scale to keep pace with new operational practices and data centre growth.

With these critical building blocks and capabilities, enterprises will be ready to leverage their data centre investments to boost performance, scalability and efficiency of their operations and gain an edge on their competition.