Networking Modernization

Networking and storage are taken for granted but their modernization is as important to the enterprise as the Applications and Databases.  When businesses embrace the cloud, they must consider whether their on-premises network will scale up to the cloud traffic. The cloud acts like a massive aggregator of traffic and with hybrid cloud, the on-premises network can get overwhelmed because they were not designed with the cloud capacity. This section of the book deals with these considerations for multi cloud adoption and hybrid computing.

Networking modernization is essential to digital transformation. When networks age, they don’t just pose a higher risk with a fault domain, they also increase complexity by being some of the lowest levels of virtualization and compute. When a single Network Interface Card failed on the corporate network associated with a production system, it was easy to diagnose given the reservations made and the stack that was dedicated to it. In a hybrid world, the customers have gone way beyond the traditional application/database landscape to having more modular applications with deep divisions and even segregated hardware. The communication is assumed to be a resource as free as the storage and one that does not factor beyond the latency of a single call. With cloud traffic, application usages and their management via a single pane of glass has elevated the customers from on-premises to the cloud. The public cloud supports rich monitoring that even spans the on-premises with the help of agents running on the enterprise hosts, but they do not help in determining the root cause of failure when the symptoms of failure become scattered, sparse, and even random or non-deterministic.

Newer networks have become software-defined and rightfully so although this has increased an abstraction layer over the hardware. This is an architectural approach to data center networking in the cloud era, bringing the flexibility and economy of software to datacenter hardware. It helps enterprise network infrastructure with the needs of application workloads by providing 1. Automated orchestration and agile provisioning, 2. Programmatic network management, 3. Application-oriented, network wide visibility, and 4. Direct integration with cloud orchestration platforms. SDN is even built into each operating system. When IT wants the ability to deploy applications quickly, SDN and network controller can be used, and policy can be managed with scripts. HyperV, and network controller can be used to create virtual Local Area Networks overlays which do not require the reassignment of IP addresses. Hybrid SDN gateways can be used to assign and manage resources independently.

There is greater security and isolation of workloads with the use of network security groups and distributed firewalls for micro-segmentation. North-South internet traffic and East-West intranet traffic can be established differently. User-defined routing can be configured with service chains that can be established with 3rd party appliances such as firewall, load balancer or content inspection. Cost is driven down by converging storage and network on Ethernet, and activating Remote Direct Memory Access (RDMA)

Network modernization might seem like an overwhelming challenge by virtue of the number of entities impacted by the effort. It can even be a struggle to get a clear picture of the evolving application environment or to document the changing requirements over the infrastructure and operations. Many organizations that don’t know where to begin can do so by identifying gaps that might hinder SDN deployment, determine automation needs, define an orchestration strategy and develop a roadmap.

A strategy for orchestration and automation becomes critical to such implementation plans. Some of these activities of network modernization include enabling self-service functions for development teams, reducing risk through integrated governance and management, preventing vendor lock-ins on hardware-based platforms, saving time by orchestrating and automating integration complexities and boosting overall quality through intelligent and aware operations such as self-healing.