Ever since compute and storage elements became virtualized, network has been a major impediment to overall data center agility. Whether we address security, automate provisioning or move applications in a multi cloud environment, the result is a tug-of-war at the network layer, and critical goals for IT are at odds with one another. Applications with immersive experience, virtual reality and huge increase in the number of IoT devices is causing an explosion in technology, setting up the network to become high performing, sophisticated and self-aware.
Conventional Network Challenges
- Proprietary hardware limits gains from industry wide-innovation
- Inability to scale up in line with increasing traffic demands
- High complexity with dedicated hardware/software for each function
- Difficult to apply a consistent set of policies for different network functions across infrastructure
- Error prone manual configuration and change management of network infra
- Limited security controls to stop persistent threats
We understand that the data center network strategy is really a balancing act – balancing costs, availability, flexibility, growth, and capacity – to meet business objectives. In fact, initiatives aimed at satisfying one set of requirements often have direct impacts on the ability to satisfy others. Business and IT leaders are demanding more technology innovation in the enterprise network space.
Software Defined Networks (SDN), coupled with Network Functional Virtualization (NFV), has been a disruptive formula. Starting as an academic experiment SDN has substantiated itself as a way to automate and orchestrate network state in a highly scaled hybrid data center environment. While SDN is used to automate management of physical underlay, NFV transforms the network functions into software and treats the underlay as backplane of moving packets.
Some of the key benefits of SDN and NFV are as follows:
Some of the key benefits of SDN and NFV are as follows:
- Greater Network Agility - By dynamically scaling up the network and adjusting network-wide traffic flow based on workload demands
- Improved Operational Efficiency - Through centralized management and orchestrated network provisioning in minutes
- Reduced Capital & Operational Cost - By eliminating dependencies on proprietary hardware and enabling simplified network administration and maintenance
- Simplified Network Monitoring & Troubleshooting - Through centralized real-time 360 degree visibility and analytics into what’s occurring across the entire network infrastructure
- Improved Security - By providing an isolated application environment and enforcing microsegmentation and zero-trust network security model
Both SDN and NFV deliver agility, flexibility, and programmability that align closely with the needs of increasingly critical applications. The popularity of SDN technically follows on similar lines to that of public cloud. The reasons the public cloud gained popularity included the efficiency, elasticity, and simplicity that it offers. At the same time, the harsh reality is that ‘economy of cloud’ is not proportional to the ‘economies of scale.’ Digital enterprises are looking for benefits of cloud and yet seek to retain control. SDN promises cloud-like scale and simplicity coupled with the added benefits of control and personalized network response to changing traffic needs. But can it be applied, and where to start the implementation? Market hype surrounding SDN adds more complexity and enterprises are uncertain on how to best leverage the complete SDN ecosystem. Arguably, some of the major roadblocks are attributed to architectural complexity, vendor restrictions, troublesome migration, cultural obstructions, application mobility constraints, multi-tenant analytics, DevOps collaboration, static network management, and insights.
Change is inevitable
For widespread adoption, SDN implementation has to change. The software-defined networking architecture must be reengineered to provide infrastructure hybridity, real time analytics and cognitive automation multi-layer controller, heterogeneous application workloads support, orchestrated links to compute, rapid development cycle, configure-once model, along with easy implementation and graceful migration.
- Self-aware SDN
The data center network needs to become more intelligent. Machine learning and cognitive automation technique, pave way for anticipatory network management, making the network more dynamic. The software-defined network can sense, analyze, and correlate network traffic patterns, providing insights to compare actual versus desired state. Real-time insights help in making planning and architecting decisions. This self-aware, software-defined networking enables to automate corrective actions, redefining real-time service assurance.
- Applications dynamism
Businesses and IT professionals recognize that their focus, the sum of their efforts in digital business, align to the applications. The changes in application deployment scenario is forcing organizations to increasingly look for SDN solutions that can deliver seamless application migration and support fast changing DevOps needs. Containers are defining the next wave of change. Methods like dockers, Mesos or Kubernetes are making it easier to run microservices even on a VM-based virtualized environment. And this network function virtualization is just the tip of the iceberg. Additional computing models along with VM, containers, and legacy applications are going to coexist. For workloads to run in harmony, network needs to be rewired to support this combination of different kernel technologies.
- IoT traffic steering and policy enforcement/unpredictable data growth
IoT is leading the next industrial wave and industries are leveraging sensors and M2M communication to gain competitive advantage. Gartner forecasts that 8.4 billion connected things will be in use worldwide in 2017, up 31 percent from 2016, and will reach 20.4 billion by 2020. The sheer volume, variety, and velocity of traffic that will be generated from these connected things will have immeasurable impact on data center network infrastructure. This creates the need for organizations to look for SDN solutions coupled with NFV that can dynamically scale up the network, auto steer the traffic, and perform real-time policy implementation to meet the demands of IoT data explosion.
The first cycle of SDN is over and the next leap of software-defined networking is defined by a myriad of dynamic changes brought to fore by new-age digital revolution. This new approach to software-defined network requires all teams (networking, systems, storage, operations, security, and application development) to look at any transformation holistically because there is a blur between where one responsibility starts and where it ends. Functions are moving from boxes to applications and programmability is not only programming applications but programming network functions resulting in a variety of interactions that need to be understood by any enterprise.
To future proof, this requires federation of network environment embracing architectures around hybrid cloud computing initiatives, DevOps, and containerization. Selecting the right vendor ecosystem and architecting the right solution has become more art than science. The resulting chaos is defining a new curriculum for network professionals who have to be skilled not only in erstwhile network technologies but also in programming languages, application design strategy such as microservices, and compute levers such as containerization.