Authored By: Ram Shankar Srinath, Senior Director, Software Engineering, Juniper Networks IEC
We live in exciting and disruptive times where everything is shifting to the cloud. Service providers are continuously under pressure to deliver rapid, low-latency and high-volume communication of information.
This has been especially important to enterprises which are increasingly adopting cloud-managed enterprise connectivity to reduce costs while increasing control and flexibility. Every company in the connected world has either decided on or is actively formulating, their cloud strategy and in this process, the network is taking on a new level of importance.
In the IT world, there is a lot of adoration for the cloud. Whether it’s service agility, highly automated operations, or scale-out architectures, IT network architects have been closely watching what the web-scale companies have been doing for years. And while not every practice is directly transferable to other parts of IT, it is safe to say that cloud architects have changed how businesses everywhere think about and build their networks.
Cloud adoption in networking industry isn't just about having more capable routers and switches. It should ideally address how networks are operated, with a greater emphasis on management, automation, visibility, and continuous integration. By moving to a simpler, more uniform set of building blocks, cloud architects have flipped the emphasis from networks to networking.
The future of networks will be less complex, more software-defined, easier to procure and integrate, more open, more stable, and much agiler in order to adapt to the needs of the business. Cloud adoption in networking is all about leveraging the strengths which a network has, and ultimately democratizing the cloud so that networks of all shapes and sizes will win.
The attributes of cloud infrastructure resulting in rapid growth and adoption of cloud are:
- On-demand self-service: an ability to provision computing capabilities without requiring service provider or human intervention.
- Resource pooling: Computing resources are pooled to serve multiple consumers in a multi-tenant model with different physical and virtual resources dynamically assigned according to consumer demand.
- Elasticity and Agility: capabilities can be elastically provisioned and released based on demand. The capabilities appear unlimited to the consumer.
- Availability: capabilities are available over the network and are accessible through standard mechanisms with complete location independence of capability.
- Metering: metering the resource usage providing a level of transparency to the consumer and the service provider.
All these characteristics have been the drivers for cloud adoption that enterprises are well and truly beginning to embrace.
Networking in the era of Cloud
Networking in the cloud era needs to combine the reliability and reach of carrier-grade networking with the control and usability of enterprise-grade networking resulting in what we call the cloud-grade networking. In order to deliver the fundamental attributes of cloud infrastructure, cloud-grade networking needs to adhere to the following principles:
Where does it run?
Networks must be able to run anywhere—on any software, on any hardware, in any cloud. It refers specifically to the disaggregation of the networking technology stack so that applications can run in any cloud, cloud workloads can be run on any device, and software is not locked to hardware.
By disaggregating the network stack into discrete layers connected via open interfaces, operators can fundamentally alter the economics of the network by leveraging common building blocks and increasing competition within the stack. Of course, anything that is disaggregated must eventually be integrated, and companies must have the freedom to choose between assembling individual components themselves or purchasing a set of disaggregated elements as an integrated solution.
How is it managed?
A self-managed or self-driven network combines telemetry, workflow automation, DevOps, and machine learning to create an infrastructure that is responsive, adaptive, and ultimately predictive. The journey to a self-driving future starts with today’s human-driven environments and expands to include more event-driven operations, and then layering in machine learning algorithms en route to a full self-driving experience.
Networking is complex, and many networks are brittle. Heavy-handed change processes are the only means of defence for many enterprises. The future, though, demands a pliable network that can accommodate the evolving needs of users and applications. By integrating and automating—not just within the network but across the whole of IT—networks can move from immovable to adaptable, network operations with more time focused on business-driving and revenue-generating activities.
How is it secured?
Security through a software-defined secure network
The days of simply securing the perimeter are over. Threats today are everywhere—both inside and outside the perimeter—shifting the focus of security from preventing attacks to detecting and quickly isolating them. This requires information to be centralized but enforcement distributed.
Software-Defined Secure Networking (SDSN), is the application of software to drive pervasive detection and enforcement, making every IT component an integral part of the security umbrella. Using the SDSN approach, security teams can maintain centralized policy and control while surfacing threat intelligence across the whole of the infrastructure so that it can be analyzed in real time and enforced dynamically
How are services built?
Companies— whether service providers, cloud providers, or enterprises—are deploying more than just a network. In fact, the network is merely an enabler for network services and applications. Every element within that network must ultimately be a platform: hardware is a platform for software; software is a platform for network functions; the network is a platform for services, and the cloud is a platform for applications.
The very nature of platforms is that they need to be designed explicitly for extensibility and longevity. A platform without anything built on top is not useful, and a platform that exists for only a moment is not practical. As a foundational element, platforms must be easily built upon, and they must be enduring. Building a corporate network demands that the underlying network is a solid platform.
The migration from enterprise and carrier-grade networks to Cloud-Grade Networking involves not just products, but also people, processes, and tools. It’s as much about first principles for cloud design as it is about the specific technologies that underpin the cloud. And to a large extent, those tools and best practices in building and maintaining Cloud-Grade Networks are immediately applicable to environments of varying forms and size.