26 Dec

Supply chain woes forcing more workloads to the cloud

    Chip shortages are driving enterprises to shift applications to the cloud, avoid ad-hoc hardware purchases, and work to improve their equipment forecasting capabilities.

    Messaging services vendor Interop Technologies runs three data centers to provide services to customers and to run its own back-office systems. Interop also provides turnkey hardware/software solutions that run at customer sites. Pandemic-related hardware shortages, particularly those of servers and storage, have put a severe crimp in the way it does business.

    “When you go to procurement, you get so much push-back,” said Joshua Collazo, the company’s director of infrastructure. “This is back ordered, that is back ordered.”

    Before the pandemic, the company was able to jump on opportunities quickly. “That’s gone away,” he said. “Ad-hoc has gone the way of the dodo for us.”

    Interop is used to adjusting for seasonal supply-chain disruptions, particularly around the end of the year. But now, more of these dead windows have appeared.

    The larger the order, the bigger the problems. It might take about a month to provision a smaller system for companies that only need a handful of boxes, but “if you want 20, 30, or 50 boxes, it looks like six months,” he said. “When you consider that you’ve got projects that could span nine to 18 months to deliver, and you’re adding six months right off the bat, it makes things challenging.”

    Shifting workloads to the cloud

    Interop has considered moving its core services out of its private data centers and into the cloud, but it’s just not feasible

    Cloud providers aren’t optimized for messaging services, and running Interop’s infrastructure on-prem gives the company greater flexibility in the services it can offer, Collazo said. “If you’re going to control the whole stack, you get a few more options than, say, Amazon’s load balancing. They’re going to do it their way, and you have to tailor everything to the capabilities of the cloud provider.”

    “We’re looking at significant refactoring to put systems into the public cloud.”

    So, to deal with the supply chain constraints, Interop has been trying to get as much use out of its data centers as it can.

    “We’re working to be as agile and flexible as possible, and leveraging key technologies like Kubernetes,” Collazo said. The company has also deployed an adaptive cloud fabric from Pluribus Networks to make its various data centers feel like a single platform to its legacy systems. That allows the company to maximize usage without having to rewrite applications that were designed to run inside a single data center.

    But the biggest factor in freeing up data center capacity for revenue-generating, customer-serving core systems? Moving the back office to the cloud.

    Some systems were already due for an upgrade, Collazo said, and moving those to the cloud instead killed two birds with one stone. In addition, some back-office systems would have required significant changes to adjust to the large number of employees who switched to remote work setups due to the COVID-19 pandemic.

    Moving those back-office applications to the cloud allowed Interop to upgrade its apps and get access to capacity on demand, which “we know we’re going to need because we know everybody’s working from home,” Collazo said. (He declined to name the specific back-office systems that were moved to the cloud for security reasons.)

    Demand from hyperscalers intensifies supply-chain issues

    Interop isn’t alone in moving workloads to the cloud in response to pandemic-related hardware shortages.

    According to newly released research from Insight and IDG, 91% of IT decision makers say that they are being impacted by IT supply chain disruption. In response, 44% say that they plan to shift processing to the cloud to reduce disruption. In addition, 43% say that they are avoiding last-minute, ad-hoc purchases, and 42% are improving their forecasting capabilities to get more long-term visibility into their equipment needs.

    Megan Amdahl, Insight’s senior vice president of partner alliances and operations, predicts that supply chain issues will continue well into 2022. And it’s not just chips.

    “I’ve heard from suppliers not being able to meet demands for just plain steel,” said Isaac Gould, global technology expert at Miami-based Nucleus Research. “And for wiring. And motherboards and graphics cards all require a lot of semiconductors. Pretty much all the components to build electronics these days are facing some sort of constraints. Not only shortages but also increases in demand as a result of the pandemic and the strain on transportation.”

    The hyperscale cloud providers have the clout to get first dibs on everything that is produced, Gould said.

    “The hyperscalers are getting the bulk of supply,” he said. “They’re opening new data centers as we speak. They’re talking about building underground data centers. They’re not slowing down at all. It’s the little guys who are facing chip shortages – not because there’s a lack of chips, but because they’re being used to build data centers for the big guys, for Amazon, for Google, for Microsoft, for Oracle.”

    “The hyperscalers do seem to have the scale advantage,” confirmed Tim Rehac, Ernst & Young Americas cloud infrastructure and strategy leader. “I haven’t heard of any hyperscalers being dramatically short on hardware.”

    As a result, the move to the cloud, already proceeding quickly before the pandemic, has greatly accelerated.

    “Before the pandemic, I would have told you that it couldn’t accelerate any more – but it did,” Rehac says. “Now, almost everything new is starting in the cloud.”

    Enterprises rethink legacy workloads

    New workloads aren’t the only ones destined for the cloud. Companies are moving legacy systems off premises, as well.

    Say, for example, a company is running on-prem applications that need updating, or need to scale, Rehac said. “Absent supply chain issues, they would have ordered more hardware,” he said. “Now they’re at least looking, or selecting, a move to the cloud instead.”

    Latency requirements and security concerns may keep certain applications on-prem, however. Or, legacy systems may work just fine, and there’s no urgent need to move them – even as hardware ages. “We’ve seen clients really sweat their assets, sometimes resorting to eBay to buy NIC cards or other parts for older servers to keep them running,” Rehac says.

    Other companies are dealing with the supply chain issues by delaying planned hardware upgrades, said Michael Vovk, managing director at Deloitte Consulting. Instead, they’re spending more money on vendors who provide extended maintenance for no-longer-supported technologies. “We are seeing companies extend their technical debt,” which can create challenges for them, Vovk said.

    Clients that are close to capacity limits are feeling supply-chain pressures particularly urgently, Rehac said.

    “A few clients are finding other workloads to move out of the data center to free up capacity in the data center for expansion of their current workloads,” he says. “We’re definitely seeing that.”

    Share this