Since its inception in 2012, the Raspberry Pi has been in the arsenal of many a developer, enthusiast and even school curriculums – but I for one have never seen it as anything more – something to play around at home with. But have VMware with their ESXi on ARM fling, opened the doors to any number of commericial applications?
On the 28th May this year, the Raspberry Pi foundation released their latest iteration, the Raspberry Pi 4, Model B. So what is special about that? Well, there’s some cool stuff, dual 4k monitor support thanks to a more powerful video module, more USB ports (2x 2.0, 2x 3.0), a slightly fast quad core 1.5Ghz processor which will not doubt excite the home enthusiast – however for the first time, there was an option of 8Gb RAM onboard.
Who cares? Well, I do. Because 8Gb is probably the tipping point for the Pi becoming particularly interesting as far as what can run on the device. No I don’t mean we can run doom smoother, or what not – thats been done. But what is interesting is a little known product that has been around a while called ESXi. Yes, those pesky VMware folks have been meddling yet again.
For you new fangled AWS-only folk, ESXi is VMware hypervisor, and drives probably 99% of private cloud and on premise workloads (no I have no proof of this, but it seems a sensible guess!). Most hardware running ESXi nowadays will have 100s GBs of RAM, or even TBs, and will be at least a single U in size, and pull the best part of 1000W of power. Which in all fairness – is great…. for a datacenter running your ‘cloud’ workloads. But what if you have a use case for something that runs elsewhere? Let call this edge compute, cause hey – nobody has coined that term!
Now before everyone moans at me, the VMware Fling in question isn’t built specifically for Rasberry Pi’s – its for anything running an ARM processor, so that could be the Qualcomm Dragonboard amongst others. But what the clever folk have managed to do, is to rewritea product heavily engrained in the x86 architecture and get a working product on ARM. With the release of the latest Pi, there are suddenly a number of very exciting possibilities.
Some useful reading…
Before I go into overdrive, I thought I’d provide a list of handy-dandy links to some fantastic articles and ideas using ESXi on Pi that I’ve seen over the past week or so.
ESXi on Raspberry Pi – Tutorial (Nico Vibert)
How to run Raspberry Pi OS as a VM on ESXi-Arm (William Lam)
Making sure my @esxi_arm on @Raspberry_Pi Host/Cluster/VMs are running smooth by monitoring with @vRealizeOps 8.2. This has to be the best @vmwflings ever released by @VMware. Everything working awesome. Next test is to create new ARM VMs with @vRealizeAuto. pic.twitter.com/RLBtXgpkyB— Dale Hassinger (@dalehassinger) October 11, 2020
So thats how to do it, and how people are beginning to build vSAN clusters, even running ESXi on the Nintendo Switch (which runs an ARM Cortex CPU). Some people are saying this is great for a little bit of fun, whether that be to run some funky software firewalls VMs, or even as a witness node for their homelabs but this made it onto the stage at VMworld for a reason.
That reason, is pretty simple. IOT and edge computing is something that is expected to grow hugely over the next few years – and having a small device with a reasonable amount of compute, completely customisable with your own VMs or containers makes this an interesting proposition. So where could this be used?
Where can this be used?
Lets think about this around our predicament across the world right now. We need to ensure the safety of the public and our employees with the ongoing pandemic, so that Pi could act as a way of analysing thermal images to alert or reject access to people with high tempertures.
Mobile vehicles such as buses, taxis, or trains could provide on-the-fly advertising targeted at those people using their service, using the edge compute to identify who is on the bus, and use 5G to download the relevant content for that person. Or even analyse mass data on a vehicle, whether that be from an ECU or a camera to assess risk without having to have that risk assessed (at slow speed) in a cloud datacenter somewhere.
Put one on a drone to assess what the drone sees in real time to assess issues with your industrial equipment.
Whever low latency, compute is required – the Raspberry Pi, in conjunction with ESXi can provide that. It opens up plenty of options for developers to write simple applications to be ran on low-resource VMs or containers to save backhauling data over expensive, unreliable 4G or 5G networks just to process that data in a cloud environment and then send an action back to that same device. Edge computing can provide real value to certain industry verticals.
Does it have commercial potential?
The interesting part is waiting to see who commerialises this. Whether that be a Edge IaaS (just ESXi on a ARM computer), or maybe even a no/low-code app built on ARM. I wouldn’t think VMware are the only business thinking of this – I for one wouldn’t be surprised to see AWS offer Outposts ARM – especially with their recent moves stepping out of their datacenters and their continued alliance with VMware. The obvious company is Dell (who own VMware) whose core business is hardware – could they produce an ESXi on ARM modular solution with bolt ons maybe? You want a Pi running ESXi with a 4G card and a GPS module, here you go. You want a Pi cluster running ESXi alongside some SSD storage for edge compute? Here you go! The possibilities are endless.
This is a very interesting development from where I’m sitting, and something that I will be keeping an eye on over the next few months to see where this technology can go in the business world!