Converged infrastructures vs SDN: integrate or disaggregate?
By Wilson Lai, Technical Director, Asia-Pacific and Japan, Extreme Networks
Monday, 06 July, 2015
Choosing the right networking infrastructure means reconciling contradictory technology trends.
Static architectures are ill-suited for today’s highly mobile and virtualised environments. The proliferation of data and growing number of devices requiring network support requires networking to be agile; thus, organisations will be looking to leverage network infrastructure agility to deliver rapid responses that support business growth.
Network agility enables applications and services to be added, removed or adjusted promptly. Yet organisations are faced with the inherent decision of determining what should be outsourced and what should be insourced. Although costs will be an underlying factor, the need to separate investment into parts of the infrastructure that are strategic to the business, while outsourcing the rest, will be crucial.
With Gartner projecting worldwide IT spending for 2015 to increase by 2.4% over 2014 to US$3.8 trillion, how will new applications, users and services be removed or merged within the organisation’s own IT infrastructure?
By now, it has been established that the network infrastructure must be flexible, rather than static, to keep pace with rapidly changing markets. So the question organisations should be seeking to answer is: do you bring the pieces closer together by deploying converged infrastructure, or disaggregate and have more moving pieces by deploying software-defined networking (SDN)?
Challenges
The accelerating IoT wave is seeing a deluge of connected devices accessing the network, resulting in a vast influx of data, transactions and users. By 2020, the number of connected things will have surged exponentially to an estimated 40 billion. This increase in connected devices will have a huge impact on organisations’ strategies for the data centre and network infrastructure. Data can become unstructured, leaving a percentage of very insightful, high-quality data that is not stored anywhere.
Data centres will need to become the modern-day railroad - in other words, be able to bridge the movement and storage of data. With existing network complexities, further network growth and expansion may pose significant challenges for management and provisioning. Once more, organisations will encounter a recurring concern - why should data be moved to a new platform, to a point where it might be impossible to leave that platform?
The answer might be simpler than we think. The volume of data flooding a network as result of IoT will need a place to be stored, exposing the traditional network problems - many are not agile enough, much less ready to support the flow of data or the pace of the ever-changing business landscape. This situation will encourage organisations to recognise the issues with traditional networking.
For a start, it is difficult and time-consuming to change and adapt infrastructure due to network complexity. The same can be said about the constant implementation of new technology features that require time before end-user benefits are realised.
Furthermore, network intelligence tends to be too siloed from applications, preventing organisations from benefiting from the presence of valuable and relevant data in the respective systems. Networking hardware is too often closed or static, limiting future scaling or enhancements of the data centre.
The network should be an asset that drives an organisation’s innovation forward, yet too often it holds other efforts back or slows them down. How can organisations address these issues between the differing trends of deploying converged infrastructure or SDN?
Closed integration
Deploying converged infrastructure removes a layer of complexity when data centre configuration is abstracted, by preassembling the components so organisations don’t have to deal with the pieces. This ensures that tighter integrations are built, along with preconfiguring of the pieces - storage, networks and computer servers - with different applications and uses in mind.
With the pieces being integrated and working collectively at optimal capacity, organisations are able to streamline the day-to-day management of their data centre. They may also stand to benefit from reduced costs that would otherwise be spent on single-use components that are required in managing or troubleshooting these pieces. By consolidating resources and outsourcing the network complexity, time taken to build and scale the data centres is reduced significantly.
Open disaggregation with SDN
Then there is SDN, which, in most architectures, disaggregates networking and IT intelligence into separate pieces in order to create more layers of abstraction and increase agility and levels of control and automation. SDN enables the option of separating networking intelligence from networking hardware, which cannot be achieved with traditional routers and switches that have mutually exclusive software embedded in the hardware.
Since the software will not be bound to a networking operating system or control plane, opportunities to leverage new innovations faster will arise. The most extreme incarnation of the SDN model has the potential to not only add complexity, but also to increase agility and choice, thus addressing many of the challenges listed above.
To the uninitiated, these may seem like opposing approaches: integrate and bring the pieces closer together, or disaggregate and have more moving pieces. Yet there are different advantages in the deployment of each trend.
Enterprises interested in getting systems up and running quickly may go to the most converged infrastructure available, while those interested in open and dynamic networks that can leverage innovations faster will look to SDN. At face value these are conflicting trends, yet ultimately it boils down to the different needs and approaches of an organisation, all of which seek to achieve the goal of greater business agility.
Cyber resilience: key takeaways from a global IT outage
One of the industry's largest IT outages in recent memory was an event that could easily have...
From a single update to global chaos: lessons from the CrowdStrike outage
In the wake of the CrowdStrike outage, IT leaders must examine the lessons learned and develop...
Walking the talk on AI integration
Successful AI implementation will determine whether it can truly deliver on its promise.