Coronavirus: Hardening up datacentre equipment supply chains post-Covid-19

Supply issues could continue

Jennifer Cooke, research director for cloud to edge datacentre trends and strategies at IDC, tells Computer Weekly that supply could yet be delayed in the next few months, although it seems that larger, multi-tenant datacentres, which may find sourcing easier, can still get what they need.

“What datacentres are seeing, however, is a spike in demand for remote monitoring tech,” says Cooke. “Colocation providers that invested in these platforms are seeing customers log in and stay logged in for a lot longer, relying on remote monitoring tools when being there in person is difficult or impossible.”

IDC analysis so far notes that while most supply chain organisations have already activated business continuity plans, these have mostly been designed for short-term, localised disruption. This suggests that business continuity planning could have to ramp up, targeting better visibility of supply chain capabilities at both ends, as well as the overall risk backdrop – making short-term adjustments where possible, and acquiring more external data through third parties.

Alternative supply sources should be located to guard against future production shutdowns, logistical constraints or custom disruptions. This might be about developing surge capacity and alternative transport options, for example sea instead of air, or substituting products – running scenarios where possible to identify potential pain points. Luckily, operators massively expanded overall datacentre capacity between 2017 and 2019, and sufficiently optimised space, power and central connectivity has been available so far.

That said, IDC reckons it could take a global effort to mitigate delays to datacentre construction. Facilities and last-mile bandwidth support to all end locations should be scrutinised. On the other hand, Covid-19 could accelerate the shift to service-provider built or operated colocation and cloud facilities, with overall power and capacity expected to expand by 8-10% over the next five years.

Monitoring amid social distancing

Andy Lawrence, executive director of research at the Uptime Institute, highlights ongoing skill shortages amid a complex picture around cloud migrations and risk perceptions that could negatively affect growth forecasts. Dealing with these issues will be key to successful future-proofing of the datacentre equipment supply chain.

“Covid-19’s impact has been about the things you would expect – dividing people into shifts, staff shortages, people having to be off work for self-isolation – but also deferring maintenance, which has caused a lot of worries,” says Lawrence. “Even missing one service on a generator can affect your warranties, and even your permit.

“Industry is moving slowly that way anyway, but everyone we’ve spoken to says they’ll do more remote monitoring in future. One long-term change is having fewer people on site, instead of visitors all the time, all different companies, all there every day – a real issue in this crisis.”

Yet datacentres are expected to continue in an expansionary mode, for suppliers that get the situation in hand by sourcing appropriately amended, updated documentation and advice on maintenance delays.

“Everyone we’ve spoken to says they’ll do more remote monitoring in future”

With more automation, remote monitoring and condition-based monitoring, unexpected failures should become less likely. Instead of quarterly technician visits, say, remote sensors and monitors could be checking data against an analytics program, perhaps with artificial intelligence, that reveals the likelihood of failure over the next 90 days.

As long as the security strategies keep pace in a more remote, automated, cloud-based world, datacentre operations could be well positioned for a post-Covid-19 future, says Lawrence.

“Datacentre management is religious about this kind of continuity,” he adds. “The level of thinking they do is extraordinary – they really take a forensic engineering kind of view of every problem.

“I’ve been to datacentres where they have rooms with beds in, two weeks’ food supply and that kind of thing. They are ready for fires, floods and famine, because what they’re paid for is to keep the thing running at all times.

“They keep critical parts on site, they usually have contracts that enable the key components to be delivered fast, and obviously they have disaster recovery plans to move workloads to another site in extremis. No one has ever seen an issue like this before, so they had to think on the fly, but they’ve done a good job on the whole.”

Dual sourcing to stage comeback

Also, says Lawrence, some of the very large hyperscale datacentres, building as quickly and cheaply as possible with just-in-time manufacturing techniques, are talking about reintroducing dual sourcing for things like IT facilities and some spares.

This practice had been gradually abandoned over the last 20 years in favour of very tight supplier relationships, but it could potentially reduce the risks that come with over-reliance on certain suppliers or segments of a supply chain. Might this be good news for some of the smaller suppliers out there?

“Yes, but what you hear people talking about and what people actually end up doing can differ,” says Lawrence. “It all has costs involved. Once the emergency is over, quite often you might get that they’re not going to make that change they’ve thought about. It’s cheaper to have everything identical – but they are quite wary of that.”

Standardisation in the datacentre has been talked about for decades, he says, but has mostly been around the IT side, for racks, servers, and so on. When it comes to instrumentation, UPS (uninterruptible power supply) or integration of software components, cooling systems or building management software, suppliers haven’t really wanted to open up.

“They’re quite proprietary,” says Lawrence. “It usually takes either disruptive new suppliers or very powerful buyers to drive change, because they tend to go their own way and they tell the suppliers what they want. So if that happens, it will probably come through the likes of Google or Amazon.”

He agrees that having more remote working has largely worked well, not least by smoothing out daily workload patterns that previously saw large cyclical peaks and troughs of demand. But new risks must be accounted for and customers assured of service delivery.

This means service-level agreements (SLAs) and contracts should all be looked at again, and with the customer’s requirements more in mind, rather than the defence of the service provider. Non-critical systems could become “must-haves” as a result.

“An example might be a ticketing system at an airline,” says Lawrence. “It wasn’t built to be 24/7 critical, but now if you can’t use it, nothing happens. People will start to look at the end-to-end risk of that system. They may need to demonstrate that more. They won’t want people coming in all the time to check.

“So, definitely, everyone needs to be looking at staffing issues. Think about operating with two teams, like at a nuclear plant, with sophisticated processes for changing from one team to another, and being able to do that for ever.”

View from the coalface

Suppliers may be reluctant to provide comment that could be construed as guidance. However, Gabriel Bonilha, Europe, Middle East and Africa (EMEA) professional services manager at datacentre infrastructure provider Vertiv, largely agrees with Lawrence on many points.

He says: “The global supply chain has focused a lot on efficiency, but now we must pair it with very high levels of resiliency. We all want the best quality, price and lead times – not much news there. These challenges will ease after the pandemic, but how long will it take? And what about the next crisis to come along? No one can say.”

Considering where to localise manufacturing and assembly might help, along with continuous improvement across critical infrastructure servicing and maintenance, expanding and optimising to keep up with demand. The industry will need to do better around spares, logistics and dependencies – even though that is likely to require heavier investment in some areas. With that in mind, Bonilha notes that the current pandemic is, in fact, an opportunity to strike the right balance between resiliency and efficiency.

Volker Ludwig, senior EMEA datacentres vice-president at tech services provider NTT, says his firm has not seen “long delays” in terms of infrastructure component delivery due to Covid-19 in EMEA. However, because datacentres are the “backbone of digitisation”, he says it is crucial that long lead times and expected build-out trajectories are well planned. Teams or partners must be able to reach the locations in question, spare parts and maintenance must available and adaptability built in.

“To manage risk, we work with a minimum of two, or ideally three, different suppliers per critical infrastructure component, such as generators, UPS and chillers, to make sure we are not dependent,” says Ludwig.

“Covid-19 has also accelerated digitisation in terms of how datacentre operators work. Where there were lots of people on-site, in particular for testing in commissioning, we now have very few staff present. Most are working remotely and participating through video.”

Paul Hohnsbeen, vice-president for EMEA IBX operations at global interconnection and datacentre firm Equinix, echoes many of these sentiments.

“Covid-19 has accelerated digitisation in terms of how datacentre operators work”

He reiterates that the pandemic has piled on the pressure, accelerating digital trends from remote working to virtual events, online streaming and purchasing. All these must be underpinned by shoring up critical datacentre operations, infrastructure and services, spurring change to working protocols and “creative planning” to keep operations going, he says. Support services need to be redesigned not only for sudden capacity expansion, but to guide customers managing their digital infrastructure.

“Datacentre operators also need a plan of action for when customers readdress levels of usage as the world returns to a new kind of normal,” says Hohnsbeen. “But what that will look like is yet to be seen.

“Support and advice from regional and national agencies have proven invaluable and we have also seen increased amounts of useful information being freely shared by industry bodies and other datacentre operators. We must continue to share knowledge to assist with the new set of challenges that customers will face during the recovery phase.”

Mark Daly, director at datacentre services provider Digital Realty, says the industry should benefit from “valuable geopolitical learnings” about equipment sourcing and location in the short, medium and longer term – and he suggests that embedding sustainability and “greener” sourcing into specifications should become more top of mind.

Adapt, but stay flexible

“Sourcing critical items locally instead of in a lower-cost country may become more popular,” says Daly. “At the same time, the importance of having multiple sources for critical items has been highlighted.

“Post-Covid-19, a company not managing its supply chain will see longer lead times. Equally, multiple parties could be competing for the same resources, either from an equipment or resource perspective.”

Daly points out that many datacentre organisations could be ramping up at the same time, with similar pressures from deferred activity. “Working closely with all parties in advance of that ramp-up will be incredibly important,” he says.

Beyond the world of datacentres specifically, the Hackett Group has been looking closely at coronavirus impacts on business. Any supply chain company post-Covid-19, the consultancy suggests, should, in the near term, optimise the design of its supply networks, as well as identify alternative supply scenarios incorporating multi-year cost and capacity modelling.

Hackett also recommends more focus on common, interoperable platforms or technologies that can easily be adopted for different use cases or locations.

“Automate core processes, including order management, planning and scheduling; standardise processes across geographies,” it says.

Final configurations should be delayed where possible, with outsourcing options ready to step into the breach, says the consultancy. Relationships should be analysed and managed to understand and enable rapid changes in business demand, with longer-term plans for different scenarios developed, even in the face of uncertainty.

But Hackett adds a caveat that is equally valid for datacentre operators: “No battle plan survives contact with the enemy.”