michelangelus - Fotolia

SaaS provider rejects “costly” VMware for Nutanix Acropolis hypervisor

Data services provider Pi rejects VMware as too costly and complex, opting for Nutanix’s computing and storage Acropolis hypervisor

London-based data services company Pi has deployed hyper-converged storage from Nutanix with the supplier’s own Acropolis hypervisor, after rejecting the virtual machine (VM) environment as too costly and complex.

Pi delivers data analytics services, including in real time, via a virtual private cloud and handles client data on-site and off, with a software-as-a service (SaaS) model. It is headquartered in London and has operations in the US, Australia and South Africa.

CEO Keesup Choe said, prior to the Nutanix deployment, the company had “a classic server-based architecture with big storage devices”.

“It wasn’t scalable and data was growing at a rate that meant forklift upgrades every few months,” he said.

When Pi initially deployed Nutanix to replace the old environment, it initially used the Linux-based KVM hypervisor. “The benefits were that our guys could extend and adapt it,” said Choe.

By this point, Pi had rejected VMware as too complex and costly.

“We looked at VMware, but didn’t have it in production and rejected it,” said Choe. “There was some customisation and optimisation on Nutanix that we could have done, but were more difficult in vSphere. So we thought, if we’re going to do all that work why not do it in an open source environment?”

“There was also the cost. VMware licensing is prohibitive in terms of RAM, and we have a RAM-heavy environment.”

“So we committed to KVM before Acropolis. They’re not mutually exclusive. Acropolis is not just a hypervisor, it’s an integrated environment that can manage all resources under Nutanix.”

Hyper-V was not considered because Pi is a Unix user.

Pi’s hardware

In terms of hardware, Pi deployed a Nutanix XCP four-node 3000 series cluster at one location and a six-node 3000 series at its secondary site. Each node has a large amount of RAM – 0.5TB – plus 1.5TB of flash and 4.5TB of HDD on board. The company runs approximately 200 virtual machines.

Pi also has commodity Dell R410 servers that run KVM as compute nodes using the Nutanix clusters for storage. It also deployed the Nutanix free community version on some older Dell 510 servers as a test-dev environment.

When Pi upgraded from KVM to the Nutanix Acropolis hypervisor, it also came with the ability to use the Nutanix Prism management platform to centrally manage its deployment and to simplify data protection across its two physical locations.

A user experience similar to public cloud

Nutanix was one of the pioneers of the so-called hyper-converged infrastructure, which combines compute, storage and networking in one box. This is a trend in part inspired by the modular hyperscale architectures pioneered by web giants Google and Facebook, for example.

Other players in the market include Simplivity, Pivot3 and VMware with its EVO:Rail.

The key benefits of Pi’s new environment are that it provides an experience similar to a public cloud environment for Pi customers and is easy to manage, said Choe.

“It’s a single software-defined architecture that solves various problems and has an intuitive front end,” he said. “With Acropolis we can provide in-house most of the capabilities that Amazon Web Services or Google Cloud provides.”

The deployment of hyper-converged hardware has also reduced the datacentre footprint and power usage by 6x, said Choe.

Read more about hyper-converged storage

Read more on Virtualisation software

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Is this article for real? Did anyone even attempt to validate it's contents?
Fact Check

#1 - the title “SaaS provider rejects “costly” VMware for Nutanix Acropolis hypervisor” is completely misleading. The CEO (who apparently is also the CFO, CIO, CTO, and general computer guy) chose KVM over VMware for a perceived cost benefit and whatever incredibly complex customization that was required (BS by the way). Only LATER was KVM upgraded to Acropolis.

#2 - this gem: “There was also the cost. VMware licensing is prohibitive in terms of RAM, and we have a RAM-heavy environment.” -- was Mr. CxO himself purchasing vSphere 5.0 from 4 years ago and dealing with the "vRAM Tax"? Get with the times.

#3 - Was this article and the time put into it really worth spotlighting a company with "operations in the US, Australia and South Africa"...and a total of 10 nodes 200 VMs?? That's the size of many home labs. While hilariously appropriate scale for Acropolis, this "win" story does Nutanix no benefit. Surely they've done better than this...at least i hope so.
Wow. I'm with ibleedtech78, this article is total trash. I'm glad Pi thinks that what they have is super scalable at 200vms. Here's the twist: for service providers, it's totally not. I run thousands of VMs on nutanix based today, and trust me it is no picnic. We're longing for the day of resources we can scale as we need to, instead of being completely vendor locked to this multifunction printer model... Not to mention the administrative headache.

Great for someone who needs to do one thing and keep the whole platform really simple for one dude to click through. Bad for service providers, or anytime with growth not even between all resource types. Nutanix will say "oh just buy another box of mostly just storage/compute/whatever" - do your math, this defeats the supposed purpose of the whole system, and it doesn't perform.

The half a mil they spent would've been better on a m1000e with blades, onboard drives, a San, and enterprise plus licensing which.. Surprise.. would've given them vsan anyway for any "benefit" they wanted with the marketing hubub of data locality.