Preparing for the New IP


17 March 2015
CK Lam of Brocade

Legacy networks that were designed for an era of “best effort” services and non-mission critical applications have become inhibitors in meeting new compute requirements where virtualisation is the norm and agility is the currency of the day.

This is an era where customers interact with businesses through technology much more than face-to-face engagements - as evident from the myriad of devices and apps in use. This means networking in the data centre must become more agile, dynamic and automated.

This transformation of the network, called the New IP, is characterised by virtualised and software-based network services running on commoditised hardware. It is also driven by customer demand for open architectures and open-source technologies to avoid vendor lock-in.

The architecture for the New IP consists of four layers: Physical Underlay, Network Services (NFV), Control Service (SDN) and Orchestration.

1) Physical Underlay

The Physical Underlay is still required because virtual assets need to run on physical assets and physical assets need physical connectivity. The key consideration is that it needs to support automation inherently and respond to a variety of automation tools at the Control Services (SDN) or Orchestration layer. 

2) Services Layer

The Services Layer is where Network Functions Virtualisation (NFV) sits. NFV is revolutionising how network services are provisioned and deployed. Because NFVs run on industry-standard x86 hardware platforms instead of custom hardware from traditional network vendors, it promises true CAPEX savings. It also promises the kind of agility never possible before, because network services like routing, firewall and load balancing can be run and provisioned the way virtualised applications are.

Another important element is the need to have better visibility and analytics of the network and applications.  Today’s virtualised data centre means that with private, public and hybrid cloud, applications can reside in and be moved to any data centre anywhere in the world, creating a data centre without walls. This is where visibility and analytics will help with security, compliance and even with service level agreements (SLAs).

3) Control Layer

The Control Layer sits on top of the Services Layer and provides a centralised way of controlling network and security devices. With the Software-Defined Network (SDN) - an abstraction where business applications can dictate how the network should respond based on the application requirements – we can have a plethora of SDN applications communicating with the SDN Controller to dictate how the network should behave. All of this is managed from a central location and, with an open-sourced SDN controller like OpenDaylight, there is the assurance that there will be no vendor lock-in.

4) Orchestration Layer

At the Orchestration Layer - whether it’s with open-sourced software like OpenStack or CloudStack or other other orchestration software - server and application administrators can provision, deploy and terminate virtual compute and application resources. With NFV, the same tool can now be used to do the same thing with network services. In fact, network resources can be spun up or down automatically based on thresholds in consumption and usage determined by the business.

Preparing for the transformation

Businesses that embrace the New IP will find that they can respond to competition and their customers’ needs a lot faster. So what should companies do to prepare for this transformation?

Firstly, there is a need to have a paradigm shift in how the network has been provisioned and managed. The days of box-by-box command-line interface configuration - which tends to be reactive, will soon be over. Companies must pursue automation and simplicity in order to achieve agility.

Secondly, there has to be a plan in place to prepare the network infrastructure for virtualised network services and programmatic access. For example, companies could make sure that new network devices that form the physical underlay support OpenFlow. Or they could work with vendors and partners to develop platforms that can bridge the gap for controlling existing aging devices.

Thirdly, companies should start thinking about bringing in new skillsets around DevOps. This could be re-training of existing staff or bringing on board new people with such skills. In the meantime, they can work with vendors with  such skills to help bridge the gap through training and professional services in software development and scripting.

Dynamic new market opportunities

The New IP creates new opportunities not just for vendors but also for companies deploying it. The agility with which businesses can move in response to market conditions will not only help with customer retention, but attract new ones as well.

The transformation has already started with many businesses learning and testing such solutions, and companies who chose to stand still will be left behind.

  • Lam Chee Keong is director of Data Center Fabric and Virtualization for Asia-Pacific, Brocade.