While technology is constantly evolving, the very nature of IT — serving as the foundation for businesses — demands that change be meted out in consumable increments. So we tend to see step-function change about once every generation, followed by incremental improvements until the next leap forward.
But what triggers these big change events to take us to IT’s next-generation platform?
IT’s catalyst for change
The dawn of IT was dominated by mainframes. These single systems included the major IT pillars: compute, storage, networking, and applications. When the Internet was born, it triggered both a distribution and explosion of users. Application architectures were forced to adapt, and the underlying infrastructure was quick to follow.
Infrastructure generally must morph to fit the applications it serves. So when we see major changes in applications, brought on by an evolution of user-side demands, we can predict that infrastructure will change as well.
With that in mind, it seems obvious that we are in the fairly early stages of another tectonic-type shift. The emergence of Big Data portends a more general rise of data-driven applications. Much like the Internet gave us more users across more space, Big Data is giving us more data across more resources. To make use of all that data, applications are evolving from the tiered application architectures that dominated the best-of-breed era to flatter, horizontally-scaled architectures.
And if history is any lesson, the infrastructure will undergo its own transformation in support.
Next-generation application requirements
If the Third Platform era is characterized by a new generation of applications, understanding what those applications require will determine how infrastructure must evolve to support those applications.
Third Era applications have the following properties:
- Horizontally-scaled – Applications will tend to be based on scale-out architectures
- Agile – With an eye towards facilitating service management, interactions (from provisioning to troubleshooting) will be highly automated across infrastructure silos
- Integrated – To achieve the performance and scale required, compute, storage, networking, and the application will all be integrated
- Resilient – Distributed applications will not be designed for infrastructure uptime, but rather for overall application resiliency (fault tolerant, not fault free)
- Security – With data underpinning many of these applications, security and compliance — along with auditability — will be key
These properties will determine how each of compute, storage, and networking must evolve.
Network’s role in this new era
If newer generation applications are oriented around the data on which they feed, the interconnect that allows those resources to communicate becomes more critical than ever. That interconnect — the network — will have to evolve to deliver the requirements for these data-oriented applications.
- Scale-out networking – It almost goes without saying that the network must scale to support the increased demand from applications driving more east-west traffic. But the requirement for horizontally-scaled infrastructure goes beyond scale-out architectures. Indeed, the act of scaling out is more important than the final state of scaled out. Applications are going to grow in fits and spurts as Big Data works its way through its infant stages. Capacity will not be added uniformly, and it will be difficult to predict exactly where and when new network capacity will be needed. Network architectures need to be easily scalable, allowing architects to add capacity in small increments as and when it is needed. Hard limits in architectures (the requirement for additional layers at certain scale points) are prohibitively expensive and time consuming to execute. Newer generation networks cannot have these architectural break points.
- Agile operations – If applications and data are more fluid, infrastructure will need to be more responsive and faster to adapt. Accordingly, networks will operate under the mantra: Automate what you can, simplify everything else. The currency of automation is data. To be automated, data must be shared between systems in programmatic ways. That integration must be a foundational part of the workflows that drives IT. But not everything can be automated. More broadly, networking will need to evolve from management-by-knob to implicit behavior driven by policy. This means more emphasis on capability driven by abstraction, rather than capability by protocol knob after protocol knob.
- Integrated infrastructure – The next generation of IT platforms will see a blurring of lines between compute, storage, networking, and applications. IT elements must work in concert to satisfy application workloads, and that will require a deeper level of integration than exists today. It’s more than devices working alongside each other; data-oriented apps will require components working together. Integration across silos will put more emphasis on application abstractions and shared data models so that all parts of the solution can act on the same underlying set of driving imperatives.
- Resilience – In a distributed world, the most important thing is avoiding resource stranding. If the interconnect goes down, data is unreachable and the applications simply cannot function. This means the network has to be more resilient than ever. But next-generation resilience is not this-generation high availability. Resilience is less about avoiding failures and more about mitigating them when they inevitably occur. The next-generation network will be tolerant of faults, using multiple paths between resources to ensure nothing is ever stranded.
- Security – The notion of security is a data-oriented world change. Security is less about devices and more about data, which means the network itself will need to be capable of isolating data where necessary and demonstrating, in the affirmative, that data and applications are secure. For businesses to take advantage of data-driven applications, they have to not only provide secure operations, but also successfully hold up under audit.
Architectures, not technologies
Step-function changes like these are really architectural in nature. Where we tend to evolve IT feature by feature during the stretches between transformational changes, the migration to new generations of applications can be more disruptive. That said, careful planning during natural expansion of current-generation infrastructure can make the eventual migration to more effective data-oriented architectures much simpler. The challenge for architects today is expanding evaluation criteria carefully to make sure that current problems are solved in a way that is consistent with the eventual end state.