NHS procurement

Part of the issue here is NHS procurement, which means IT teams must evaluate the market, ask for the release of funds and then have a limited period during which they can finalise the purchase. 

That means, if prices fluctuate wildly, things can go wrong, said Ciezak.

“When Nutanix quote, it is based on today’s date and what the dollar is worth,” he said. “And then we have that for 15 or 30 days, but you cannot get everything done within 30 days, especially when your organisation and your executive has to approve it.

“It has to go to the ICS [local integrated care system] to be approved by them,” said Ciezak. “And then this needs to be approved by NHS England, and they only meet once a month. So, before you get everything done, they say, ‘Oh, now the price changed’.

“And right now, we do not know what is actually going on with the dollar.”

How will things be resolved?

Ciezak said: “I’m waiting for what will happen in the next one and a half years, when things become stable in global terms. We do not know whether that will happen. And right now, I know that people that went to the cloud five years ago, when it was relatively new, now they’re coming back because they cannot afford it.” 

PAH’s original move to Nutanix came in 2019, by which time its EMC VNX external storage was seven years old and Dell support had stopped in 2018. Since then, it ditched Microsoft Hyper-V on Nutanix and moved to the hyper-converged infrastructure supplier’s Acropolis hypervisor. 

Ciezak said he’d never heard of Nutanix or hyper-converged infrastructure (HCI) at the time, when he spoke to Computer Weekly at Nutanix’s .Next 2025 event in the US last week.

HCI saw compute and storage bundled together in nodes that could connect in grid-like fashion to form clusters, often with server and storage components scalable independently. This was a particularly attractive proposition to customers that lacked deep skillsets as they were relatively easily deployable and scalable.

“We went to the market to see what was out there and at the time we never knew who Nutanix was,” he said. “Also, we were bombarded by the big vendors as soon as I mentioned the project on LinkedIn. So, we had everything lined up to go with one of the big boys, but I had a deadline of three weeks, otherwise I’d lose the money allocated [due to procurement system constraints]. Nutanix said they could deliver within three weeks, and they did.”

And, according to Ciezak, the trust saved big by going with Nutanix and hyper-converged instead of a traditional storage supplier. “We saved nearly half a million because the other vendors came in at around £1m while we paid around £550,000 for Nutanix,” he said.

That was in addition to savings of around 75% on power, cooling and datacentre floorspace as the EMC setup’s 30 nodes and 2x 42U of rack space was reduced to 16 Nutanix nodes. That setup now supports around 450 virtual machines and 260 different applications. 

Initially, Nutanix was configured with the Microsoft Hyper-V hypervisor because Ciezak was well versed in it, but Nutanix tried to persuade him to go with their Acropolis hypervisor (AHV). 

“Nutanix offered us AHV,” said Ciezak. “I said, ‘No, I have no idea who you are. You have to do Microsoft Hyper V, because that’s what I know. I spent 11 or 12 years studying this. Maybe later on, when I’ve learned your platform, I will consider AHV.’”

After that, some Microsoft Hyper-V functionality became unsupported in Nutanix. That prompted him to consider migrating to AHV, and that took place during the Christmas period in 2024.

The same number of Nutanix nodes remained – 16 split between two sites – but the hypervisor was migrated from Hyper-V to Nutanix’s AHV. 

The key benefits for Ciezak were latency, boot time and general ease of management. “Latency between nodes and VM guests on Hyper-V is 15 millisecond-plus. On AHV it is sub-millisecond,” he said. “Hyper-V is a hypervisor over a hypervisor. Using AHV and [Nutanix’s OS] AOS, you don’t have to go through the virtual storage protocols anymore, because you directly access the storage.

“That was the biggest gain,” added Ciezak. “To boot in Windows, when we were doing Windows updates on a monthly basis, let’s say server 2016 on Hyper-V, it took about seven minutes to do the boot. Now, it takes 15 seconds or less on AHV.”