Inside track: To disaggregate or no
“But if you’re to look at an enterprise where the network is very centred internally, then I think it has less relevance in the market.”
Appear and disappear
According to Ali, the network industry is a rapidly evolving marketplace and it is normal to see different models appear and disappear.
“The core thing that we talk about is when you have shifts in technology and disaggregation, whether it’s a good idea or not to adopt them really depends on the operating model of the organisation. These new approaches that are coming into the market may be suitable for some organisations and not for others,’ he said.
“We take a very market-driven view – what is the operating model, what is the business case and what are you trying to achieve here? If it’s pure cost reduction, it may not be the answer.”
But when you start to talk about performance optimisation, gaining control and being less at the mercy of vendors and having more ability to plan your own destiny with your own dev ops team, then Ali thinks it can start to make sense.
“Whether network disaggregation is suitable for enterprise scale companies depends quite a bit on what you consider an enterprise. For example, Facebook has reached a size where its footprint of installed infrastructure is really unprecedented in normal enterprises,” said Emir Halilovic, programme director for networking and infrastructure EMEA with IDC.
“We tend to classify it as a web-scale service provider rather than an enterprise on its own. On the other hand, you have traditional enterprises such as financial institutions which have large infrastructure footprints as well. However, they definitely do not fall into this category.”
The debate over whether disaggregated infrastructure and hyper converged or software-defined data centre infrastructure is suitable for normal enterprises can be quite heated. However, according to Halilovic, this doesn’t really reflect the reality of the situation.
“The true situation is that you have to also count how those enterprises are using the infrastructure and how they actually approach IT in general. One element that usually is neglected in such discussions is the workload and that applications are constantly migrating to cloud,” he said.
“When they migrate to cloud they actually use those web-scale provider infrastructures which are disaggregated. So in reality what is left within the enterprise realm is usually running on traditional types of infrastructure.”
Migration to cloud
However, workloads, applications and traffic are all migrating slowly but surely to cloud-based data centres, be it hybrid cloud deployments or public cloud.
“In that sense that’s where you actually get the increase in demand for desegregated types of infrastructure or hyper-converged infrastructure. So I don’t think suitability is quite an issue. If you are running a traditional IT environment with traditional applications, you want to house them on site the same way you have done it for the past decade or so,” said Halilovic.
“It doesn’t really make sense for you to go all hyper-converged and replace everything and do the whole capital expenditure thing based on that.”
Inevitably workloads are going to migrate towards hyper-converged infrastructure in somebody else’s data centre because it is less costly to acquire new software capabilities that way.
“That’s basically the gist and logic of what is going on right now with infrastructure, because to be able to realise all the advantages of software defined data centre infrastructure such as a high level of automation and so on, it requires a couple of other things that are not necessarily present in every enterprise organisation,” said Halilovic.
“Firstly, you need to have in-house software development capabilities such as dev ops, which not every large enterprise has, let alone small and medium ones. Secondly you have to completely reorganise your IT to work with desegregated infrastructure. That’s simply not something that a lot of small, medium or even traditional enterprises are capable of doing.”
From this point of view, it is easier to acquire new software or application capabilities using software-as-a-service delivered from a data centre.
Meanwhile KPMG published a Foresight white paper earlier this year focusing on emerging trends for 2016 in which it predicted the need for more sophisticated asset management, as companies seek to get more value from their investments.
“As infrastructure owners shift their focus from buying new assets to maximising the performance of the assets they already own, the need for more sophisticated asset management has risen up the agenda. Over the coming year and more, we expect to see both public and private sector owners and operators become more sophisticated stewards of their assets,” it said.
“In part, this is because owners are keen not only to achieve the full expected lifespans of their assets, but also to get more productivity out of their existing operations. The achievement of peak operational efficiency, the better management of demand and capacity, the reduction of maintenance costs and the delivery of improved customer service can enable owners to gain access to the full potential of their investments.”
“Advances in technology, including the use of data analytics, are also adding to the sophistication of asset management. No longer must decisions be made on ‘rules of thumb’; today’s available technology allows asset managers to carefully monitor their assets in real time and, in doing so, improve decision making, manage capacity and maximise efficiency. And, in the longer term, the increased use of predictive data/analytics will shift the focus from reactive to proactive maintenance.”