SDN and NFVThe evolution of the data centre

The data centre world has both evolved and come full circle from its early roots. At one time a data centre would be a central location packed with a mainframe computer and its attached storage. With the move to PC based computing, the servers were decentralised and pushed out away from the data centre into local offices. With the rise of Virtualisation, the principle of the mainframe computer once again became the model and this is embodied in the principle of cloud computing. Whilst a cloud deployment itself is formed of many discrete components, it presents a single pool of resource to the user so appears as a single mainframe system.

This approach has been refined and commoditised within the compute and storage space and large scale deployments of servers and storage topped with a virtualised workload have been commonplace for many years now. Further developments in this area were aimed at ever increasing automation around the provisioning of servers and storage to satisfy the need for servers to run applications. Features such as live migration, the ability to move a virtual server around the cloud cluster seamlessly, and automatic failover, the ability to have a redundant server spin up in the event of a failure, revolutionised the provisioning of the data centre environment and the ability of organisations to handle business continuity planning and other mission critical initiatives.

Throughout these developments, the physical network infrastructure remained as it always had: an external infrastructure constrained by wires with a largely static setup. VLANs allowed a degree of flexibility but ultimately, change in this area was slow.

The role of SDN

As a response, the virtualisation industry came up with various Software Defined Networking, or SDN, solutions that allowed the virtualisation platform to form connections over the top of the existing infrastructure and enabled the network to respond with the agility and flexibility that already existed within the virtualisation layer. Solutions such as NSX from VMware and Contrail from Juniper provide this capability and, because they are defined and operated entirely within the virtualisation layer, delivery is highly automated. Today, right now, it is possible to have a cloud network where workloads can be provisioned and immediately connected to each other directly, without worrying about the inflexibility of the fixed network below it.

The final piece of the jigsaw

The final piece of this jigsaw is NFV – network functions virtualisation. This is best described as a network capability, e.g. a firewall, running as a piece of software rather than a piece of hardware.

This is not a new idea, the earliest firewalls were simply applications running on a generic PC, but as we demanded more and more speed, the move to dedicated hardware was required to keep up. However, as computing power got faster and the mathematical techniques involved got better, this, just as with the mainframe, came full circle and now there are a number of use cases where running virtualised network capability is not only desirable, but recommended.

The data centre made portable

This presents us with an interesting opportunity. It is now possible to deploy a virtualised workload, including all the network services from routing and security through to load balancing and VPN, entirely within the confines of the virtual environment.

By doing this we gain many advantages:

  • Speed – because all of the networking is happening inside the virtual environment latency is reduced to a minimum. In a situation where a blade chassis is used, latency is about as low as it gets.
  • Agility – automation allows workloads to be increased, decreased or reconfigured on demand, even in response to automated cues within the environment itself.
  • Scale – because the architecture is now virtual, potential scale is only limited by how many servers are installed.
  • Flexibility – now the network is just information stored on a disk, it can be moved anywhere at any time

The last point here is key. It is this flexibility that leads us to the idea of the portable data centre. Virtualising components in this way means that every element that defines the data centre and its capabilities becomes just a collection of files on a disk. This can be moved around the world simply by copying it onto a USB key.

Providing the ultimate freedom for business continuity planning, this infrastructure can be built in the public cloud and migrated onto a company’s own virtualisation platform when cost, power or resources dictate. The company can then operate within their own environment and dynamically burst into the public cloud to cope with spikes in demand. The data centre infrastructure has become truly portable and can be spun up anywhere that has some servers and a layer2 switch to connect them.

Scale constrained only by our imagination

With this architecture, scale becomes a simple case of replication. Where more computing power is required, the existing infrastructure can simply be replicated and brought online wherever needed. SDN capabilities allow seamless connection between the infrastructures as if they were side by side so they can be operated as one giant distributed data centre. We can keep doing this as many times as we choose giving us scale only constrained by our imagination.

Should large scale be required within a single location, an underlying foundation architecture can be built that supports this capability and allows growth from the smallest beginnings to hyper scale without changing the base architecture.

Building the Portable Data Centre

Our portable data centre can easily be deployed using an organisation’s existing toolset as well as a number of key products from Juniper. With the addition of VMWare and F5, automation can also be used to control the deployment of the networking functions to reduce the time taken to configure and spin up services.

The virtualised portable DC can start small and scale as large as is required without changing the deployment model at any point.

For more information or advice on SDN and NFV solutions, get in touch with our HardwareSolutions team.

Alternatively, find out more about the concepts of SDN and NFV on our What is SDN? and What is NFV? pages or visit our Software-Defined Data Centre reference architecture to understand more about the deployment of SDN and NFV within the wider data centre architecture.

Follow us on LinkedIn, Twitter, Facebook and Google+ to make sure you never miss a post.

 

About HardwareSolutions

HardwareSolutions (part of the Hardware Group) is a global multi-vendor leader in advanced IT solutions, services and support. HardwareSolutions’ team of experienced and certified consultants, partner with companies to identify, implement and support advanced network infrastructures that align companies’ technological business requirements. For more information, please visit www.hardwaresolutions.com