Patricio Martelo, Alcatel Lucent APAC strategic sales talks right-sized and application fluent network data centres.
When thinking of data centres, we often think about very large enterprises or web-scale companies and try to extrapolate to the average-size enterprise.
The reality is that while the challenges faced are similar across different segments, the size of the issue at the lower end of the scale does not warrant the complexity and cost of the approaches used at the higher end.
We can still, however, capitalise on the same trends and tailor our solution and value proposition to suit the market that we want to address.
Firstly, cloud. Whether private, public or hybrid, a data centre cloud has a switching fabric as its underlying infrastructure. A fabric provides performance, resilience and storage convergence to virtual workloads.
Much of the battle amongst switching vendors has focused on data centre fabrics of tens of thousands of ports.
Such massive port counts can capture a headline but are hardly representative of the data centre market with the majority of the DCs having less than 1000 ports.
With this in mind, a research note by Gartner – Rightsizing the Data Center Network, Mark Fabbi, Caio Misticone, Gartner, 2013 – has shown the savings that can be attained by adopting a DC fabric that is based on small-form-factor switches.
Our five-second elevator pitch could be 'A rightsized cloudenabled DC fabric can save up to 70% in Capex and more than 30% in Opex'.
Secondly, SDN. This buzzword is sometimes considered to be synonymous with Openflow, VXLAN, Openstack or other emerging technologies.
Rather than focus on these, let’s focus on what SDN pursues. In most DCs today, workloads can be deployed and optimised with a simple drag and drop operation, but time-consuming processes and configuration are required on the network.
The goal of SDN is to match this flexibility and automation. But even large enterprises lack the teams of skilled engineers that it takes to evaluate, let alone deploy, SDN.
Therefore, an evolutionary rather than revolutionary approach is required to achieve the same outcomes in the enterprise space. We call this application fluency.
The application fluent network (AFN) DC bridges this gap by understanding and automatically adapting to compute resources.
With AFN, coupled with a network management system (NMS) for visibility and APIs to the Hypervisor manager for coordination, we can offer a no-nonsense approach to SDN in the enterprise today that can be expanded with Openflow, Openstack and others if and when required.
Last but not least, we can take advantage of these trends to differentiate with our business model and value added services. For instance, the 'rightsized' cloud DC trend will be best complemented with flexible PAYG models.
The SDN trend requires a new range of professional services with skills around SDN controllers and open source software and could be leveraged to develop an SDN-based solution that targets a specific vertical.