The Network’s Role in an Adaptive Data Center

In this week’s Voices of the Industry, Rajendran Avadaiappan, Chief Information Officer at Aligned Data Centers, explains the role and responsibilities of the network in an adaptive data center.

My colleagues and I talk a lot about dynamic infrastructure in terms of data center power and cooling systems – because that is where the biggest upfront cost is. But the concept applies to the network, too. Just as there are tremendous benefits in terms of cost, reliability, and flexibility when power and cooling infrastructure are adaptive, the same benefits can be realized when network infrastructure is adaptive.

One of the chief benefits of the cloud is its flexibility. Users can provision the compute, storage, and network resources they need without regard to how utilized the data center is from a power, space, or network perspective. With that flexibility, users can innovate and grow.

Rajendran Avadaiappan, Chief Information Officer, Aligned Data Centers

In the same way, the data center could be adaptable, too. Data center infrastructure (power, cooling, and network) can be designed to adapt – expand or scale – as technology and business needs change. It is dynamic infrastructure for a dynamic world – a data center with power, cooling, and network infrastructure that doesn’t have to be fully built on Day 1. It can be added or reconfigured as the needs of its users change.

Blocks within a block

From the network perspective, an adaptive data center is built with block-like network infrastructure that itself is easily scalable, within which are virtual blocks that can be seamlessly reconfigured. Building the physical network with blocks that can be as easily scaled as connecting one Lego to another means the network can flex without having to reconfigure the entire backplane, or even call in a team of network specialists, as is typically the case.

Just as there are tremendous benefits in terms of cost, reliability, and flexibility when power and cooling infrastructure are adaptive, the same benefits can be realized when network infrastructure is adaptive.

Within the physical network block, virtual blocks can be set up, modified, re-routed, all through the software portal that controls the network. So now cable management is eliminated, which reduces operating costs and improves reliability. It also reduces capital expense, because you don’t have to think ten or even five years out and build network infrastructure today for that future demand.

When a new block is added, the network auto-discovers it. It’s a self-learning, self-configuring, and self-healing network. If a port goes down on a particular switch, for example, the network doesn’t go down; it automatically re-routes traffic away from that port. In a world in which network traffic increases exponentially (welcome to the zettabyte era), a truly adaptive network enables businesses to respond to changes in technologies and in customer demands.

[clickToTweet tweet=”A truly adaptive network enables businesses to respond to changes in technologies and in customer demands.” quote=”A truly adaptive network enables businesses to respond to changes in technologies and in customer demands.”]

Growth across data centers

Being able to easily add network blocks as described above is one element of scalability – essential to the data center’s ability to adapt. Another element of scalability is the network’s ability to support more traffic through each port. Every network port needs to increase in capacity – servers are built to process more and more data, which means the ports that are associated have to grow in terms of bandwidth that they can push in and out.

There is a third element of scalability: the ability to flexibly associate or dissociate a particular server or port into a network or out of a network. Organizations are increasingly pushed to decentralized data center models to support users at the edge. But most organizations want to be able to handle all traffic across their data centers on the same logical network. That requires the network to be logically scalable beyond and across physical locations.

The challenge for most organizations in achieving that kind of scalability is not cross-state or cross-country connections but rather the “last mile” connections between the local carrier hotel and the data center. The telecom providers have invested heavily in fiber between carrier hotels, but the last mile has typically been left the responsibility of the data center provider or the individual customer. So that is where the traffic jam usually is, creating throughput challenges.

But if the data center provider owned dark fiber from the data center to the carrier hotel, that could dramatically increase scalability and reduce cost, essentially enabling customers to buy last mile transport or even flexible bandwidth as a service to support variable and need-based network traffic. In a digital economy where the fastest often wins, having a private lane on the fiber superhighway can be a tremendous competitive advantage.

Adaptive or burst

Perhaps it goes without saying, but the challenge with static network infrastructure is that it locks users in to the technology that was the latest and greatest at that time. Data center owners that built their data centers with one specific static technology are in a tough position to innovate much until that data center is depreciated. In contrast, dynamic infrastructure is designed specifically to adapt as the world adapts, to remove that burden of legacy.

Rajendran Avadaiappan is CIO at Aligned Data Centers, which provides the data center as a utility for cloud and enterprise service providers who require greater control of data center costs and faster time-to-market. Aligned Data Centers’ evolved approach mitigates the need to forecast future IT demand and provides control over capacity, so its clients can better align the data center to the needs of their business.