Azure Nears Capacity Constraints?

Surging demand for Azure's public cloud services has crashed into the global microchip shortage, according to recent media reports.

One firm told the Telegraph that they had sought to order Azure services only to be told "Unfortunately, due to high demand in this region (UK South), we are not able to approve your request at this time."

IT news site The Information reported that up to two dozen of Azure's 200 global data centres were running close to capacity, citing anonymous Microsoft employees.

Microsoft seemed to confirm there was a capacity issue of some sort, telling IT Pro: "Across the globe, we have seen unprecedented growth in the Cloud. With this surge, coupled with macro trends impacting the whole industry, we’ve taken steps to address customer increases in capacity while also expediting server deployment in our datacenters. Our priority remains ensuring business continuity for customers. In addition to managing and planning for growth, we actively load balance as needed. If it does become necessary to put capacity restrictions in place, we will first restrict trials and internal workloads to prioritize growth of existing customers."

Despite the Telegraph highlighting Azure's recent work hosting Ukrainian Government IT, that's unlikely to be the main cause of the problem. The growth in demand for Azure services is far broader than that. Microsoft's "Azure and other cloud services" revenue for Q2 2022 was up 46% on the prior year. That's revenue, not processing capacity, but it gives a hint of the "unprecedented growth" Microsoft is having to cater for. 

Azure Hasn't Run Out of Capacity

It should be stressed that Azure isn't believed to have run out of server capacity.

Cloud providers routinely provision extra capacity to accommodate future growth expected in the medium-term. Capacity is also left spare so that certain workloads can 'failover' to backup systems without overwhelming them.

Azure is just looking to safeguard some spare capacity for existing customers, while it awaits the arrival of its latest servers.

Just a Temporary Problem

Rising inflation and rising interest rates are causing consumers and businesses to tighten their belts. As a result, Gartner expects PC shipments to fall by 9.5% this year (2022). This may shorten the chip shortage.

While an impending lack of server capacity looks like a problem, it's also a sign that earlier investment is close to paying off sooner than had been expected. That makes it easier to bring forward investment in additional capacity and to upscale that investment so that capacity crunches are less likely to be repeated.  

Cloud Providers Plan B: Temporarily Delay Retirement of Older Hardware

Even if the chip shortage delays the arrival of new servers, cloud providers aren't helpless. They can temporarily postpone the retirement of older hardware.

Cloud providers typically replace their older hardware with newer models that are more power-efficient and space-efficient, as this keeps data centre costs down and minimises hardware maintainance.

If your workloads have been hosted on a virtual private cloud for four years or more, the odds are high you've already been migrated to newer hardware at least once, without noticing.

It's not the end of the world to keep older systems around for a few months longer. The old hardware isn't failing - it's just not as cheap to run as newer replacements.

The Big Public Clouds Aren't Your Only Hosting Option

If you want unrivalled scalability, per-second billing and hosting in dozens of countries, hyperscale public clouds (AWS, Azure, Google Cloud Platform) are unbeatable, ordinarily.

However, most firms don't need THAT degree of scalability. They need enough to migrate existing workloads over gradually and to grow their resources as customers sign up for their cloud-hosted service. But that's about the extent of it.

Per-second billing sounds great in theory, but it means missing out on steep discounts that come from making a longer commitment. Most hosting customers we talk to prefer to have predictable monthly bills rather than utility-style billing that results in varying monthly charges for 'compute', object storage, data egress and data backup.

Few businesses want to host their workloads in multiple countries - given the legal and compliance complexities that creates.

Hyperscale cloud providers expect their customers to interact with them via online portals, shells and API calls. If those customers need advice, there's just online documentation. This approach works well for large multinationals with a deep bench of IT talent, lots of software developers and external IT consultants. It also works well for well-funded startups creating cloud native apps from scratch.

But most firms don't fall into those two categories. Most firms need a bit of help with solution design, workload migration and ongoing platform support. That's why the hyperscale cloud providers tend to partner with consultancies that can help with such matters. However, big-name cloud consultancy services are beyond the budget of many UK firms.

Given all the above, there's no reason to worry about Azure running out of capacity. If it did - which is frankly unlikely - it would only be temporary. And there are plenty of other cloud hosting options available to UK organisations were they to be barred from ordering from Azure.