In, out, shake it all about

Written by Phil Alsop, Editor, DCS Europe Published 2018-07-02 13:49:33

When it comes to new technology, distinguishing between the hype and the subsequent reality is always a tricky judgement call. For example, a number of years ago, InfiniBand was going to be the connectivity fabric of choice – offering speeds far in excess of anything else in the market – replacing the various, piecemeal alternatives available at the time. Fast forward to the present day and InfiniBand has found a relatively small niche inside the server, and it is Ethernet that has become the dominant networking option.

At the same time as InfiniBand was ‘pulling up trees’, big things were predicted for grid computing. However, the ideas seemed to go the way of InfiniBand… except that now grid, or mesh, computing is making something of a comeback, with pervasive computing seemingly the end goal.

All of which serves to illustrate just how difficult it is to plan for changes in the data centre and the accompanying IT infrastructure. What’s going to become the next big thing, and which next big thing will very quickly become the next false dawn?

Right now, there’s incredible momentum behind a variety of technologies and ideas already having a big impact on the data centre and predicted to have an even bigger impact over time. These include: mobile IT provisioning, the Internet of Things (IoT), edge computing, content delivery networks (CDN), Big Data, Artificial Intelligence (AI), Machine Learning (ML), Robotic Process Automation (RPA), Virtual and Augmented Reality (VR/AR) and GDPR.

Let’s start with that last topic – GDPR. No need to tell anyone what it is. We’ve been living under the GDPR Cloud for months now, and recent news has only served to heighten the nervousness of those charged with data compliance. In terms of the data centre, it’s well documented that GDPR requires security and auditability. Data needs to be kept safe and the data holders need to not only guarantee data safety, as far as they can, but also to demonstrate such safety to the regulator.

However, what seems to have been forgotten in the focus on data security, is the right for individuals to be forgotten. That’s to say, if an organisation is asked to remove all traces of an individual from its database(s), then this task has to be carried out comprehensively. It’s all good and well to deleted Mr or Mrs Smith from the live database(s), but what about the backups, and the copies of the backups. And what if their details are in a tape archive. How will this information be discovered and erased?

More obviously, what I call the acronym technologies - (IoT), CDN, AI, ML and RPA - make a set of new, or at least increased, demands on an organisation’s data centre and IT infrastructure. The vast amounts of extra data they generate will need make extra demands on compute, network and storage resources and, not least in terms of where these resources are located. Speed, flexibility and resilience will be paramount when it comes to dealing with these new applications.

It may well be that such attributes will need to be delivered closer to the data generation and consumption points. In other words, large centralised data centre facilities will need to be augmented by more local, edge facilities, capable of processing data in near real-time.

While autonomous vehicles are often used to illustrate this type of edge application, more mainstream scenarios can be found in, say, the retail sector, where shop owners might well want to interact with customers as they walk through the store. Any latency (the time it takes for data to travel from the shop staff to a centralised data centre and back again to the customer) might lead to missed sales or customer service opportunities.

Performance issues are also a major concern when it comes to CDNs. The media sector is very well-versed in the importance of ensuring that content can be accessed quickly and reliably by end-users. Hence many media organisations ensure that they have not just centralised, but regional and local data centres where content can be ‘pre-loaded’, so that customers accessing the latest music or video file have a great user experience – no hint of slow download speeds, let alone buffering. And this central/local model will fluctuate as new films and music are released, and then the viewing/listening demand dwindles over time.

Ah, but with the advent of Virtual and Augmented Reality, it’s not just media companies that need to worry about optimising CDNs. More and more companies will be using virtual reality to demonstrate new products and services to their customers over the web. A great VR experience viewing a proposed kitchen layout, for example, could well lead to the customer signing up for an order; a bad VR experience and the same customer is going to look elsewhere.

Lastly, the Big Data explosion shows no sign of abating. And the ability to fire up, consume, and then stand down, the necessary data centre and IT infrastructure to run a major database query requires a constantly changing level of resource depth and flexibility.

Data centre managers need to talk to the application owners in the business, alongside the IT folks, to determine what the likely data centre/IT infrastructure workload will be over the coming months and years. Once this requirement has been understood (the fact that it will almost certainly be forever shifting), it’s time to work out what can best be provisioned internally, and what is better outsourced – whether it be to a colocation facility and/or the Cloud and/or a Managed Service Provider.

Flexibility and agility are the watchwords of the digital business. The ability to respond to new customer demands quickly and comprehensively need to be mirrored by the data centre and IT infrastructure of the business.

Finally, to flexibility and agility we need to add one more word: automation. Human beings are still required to make business decisions but, increasingly, once these decisions have been taken, automation can provision the infrastructure required to turn the decisions into reality. In the time it takes humans to understand, think and then act, automation software (AI, ML, orchestration and the like) will have long since finished several such tasks.

Related Articles