So if you have been following my blog you would know I was lucky enough to attend the Ravello Blog Day here and this was a great experience and I can now certainly say what is now GA is the same
Firstly I just wanted to summarise what this solution can do for you and in its simplest form. It can take your existing VMware VMs and run them in the cloud with no or very little changes. Again most my changes were down to bad ex coding where people referenced Public IPs and not DNS of which is a major gripe of mine. But essentially what this means is you can take your legacy system(s) and run them in the cloud now on three platforms Google, AWS and now the bare metal Oracle cloud. This is great for testing, development, patch runs and in some cases being able to break these legacy apps away so you can do VMware upgrades if OS or software isn’t supported.
They are able to do this by emulating the drivers with their proprietary technology HVX of which their summary is below
Data-center like capabilities on public cloud
Ravello enables data-center-like capabilities on Oracle Cloud Infrastructure with its next generation nested hypervisor – HVX. HVX comprises of three components:
- Nested virtualization engine – that runs the VMware VMs on underlying cloud without needing any modifications
- Networking overlay – that offers a clean Layer 2 network to the guest VMs (including broadcast & multicast capabilities) typically unsupported on public cloud
- Storage overlay – that abstracts the underlying cloud storage and exposes block devices to the VMware VMs.
HVX’s nested virtualization engine supports three modes to offer unparalleled performance when running VMware VMs on cloud. These nested virtualization modes are – hardware assisted, direct on bare metal, and software assisted.
For me the networking element is great as I have come across legacy apps that of course cant route and need that layer 2 broadcast. This is also typical also down to the fact the cloud vendor don’t want to expose this to you. The other benefit of HVX now being able to run on bare metal is the performance gains you can get as they can expose the hardware directly to your VM and also tap into those all important hardware assists to get performance gains.
What this means is you can now get 32vCPUs and a whopping 200gb of RAM per instance. So this leads me onto my experience and use cases as I was lucky enough to try and run some system on this over the last week.
For me I am going to start to look further into this but i think for me I will be seeing if I can run a test and training lab on this kind of platform for the following reasons
- Its easy to lift and shift with VMs staying on the current IP subnet
- Due to this I can simulate errors and train colleagues on how to fix issues on the live environment without affecting live
- We could make a as near live system with permissions but nullified data to train our sales guys on our products with a similar day to day experience
- I can template this whole environment and then have multiple running at once for a stream of training and testing
My initial idea was building out from Alastair Cook’s amazing system AutoLab but I wanted to put a few extra things on the edge for ease of access such as a Citrix NetScaler of which I could build some desktops or publish our app direct to the web but still require credentials to access it.
Overall the system built fast and much quicker than running this the Google or AWS infrastructure. At this point I have to say that their support team are exceptional and its a breath of fresh air that they logged into my instance as I started to get this below
Firstly I was like oh maybe I have just run out of credits but we were granted full unlimited access to this during the test phase of which I found very kind of them. What they had told me is that there are a few extra flags to check once you turn things up to 11 which is what I did. If you want get the full advantage of the system just ensure that you make sure the preferPhysicalHost is flagged as I managed to turn this off some of VMs whilst I was poking around. Again a great spot and quick response from their support team considering the timezone differences.
I am going to continue to work on my blueprint with my vExpert hours despite is being slower with the fact I know I can crank this up and test to my hearts content if required. It is also going to help me immensely with some of our apps where I forever get the blame its my hardware/infrastructure over any code or data. This way I can easily just spin it up and see if I get the same performance issues as even the networking & VMware levels are the same. The only thing I may have to do is create a VSAN like Tim has done here
So just to summarise its easy to migrate, networking is simple and you can get stellar performance
And if its good enough for the below companies its certainly worth keeping an eye on
You can also check out some of the other delegate and beta testers thoughts here and not just take my word for it