Open-Source and Adoption Efforts
Let’s be very clear that there are countless open-source projects out there and many of them are indeed great if they could only be suitable for enterprise consumption, taking into account that large enterprise are typically less receptive or adaptive to the benefits of open-source software. That perception has changed, however, and it has changed in recent years if not recent months. The proof is in the pudding, given that Arctiq is busy as ever and we are constantly looking for the next great enterprise-ready open-source solutions. I should add that this does NOT discount startups and small to medium sized businesses.. in fact it should help those be more encouraged to take the dive into open-source solutions for common AND edge-case business needs.
It seems like many businesses using open-source solutions (even as simple as RHEL for an OS, or container tech) tend to find many small-component changes when running adoption projects, such as “DevOps” efforts, or “Automation” push projects. There are other components of business IT that sometimes people are simply unaware might give an edge or even cost savings. I stumbled upon a lesser-advertised feature that was recently added to Cloudforms (Red Hat’s supported version of ManageIQ, known as the ‘hybrid-cloud single-pane-of-glass’, or ‘Manager of infra managers’) known as the Infrastructure Migration Solution (IMS), when coupled with a RHV hypervisor (cluster).
While everyone seems to be swapping out VMs for Kubernetes clusters, be it OpenShift or vanilla open-source Kubernetes, the fact that remains is many (if not most) of those clusters are NOT hardware, and still rely on VMs. The other workloads that continue to be built on VMs continue to be
- software-defined-storage systems
- heavy/ professional media encode/decode
- legacy applications that simply won’t/ can’t move to containers
- something you can think of that I haven’t included :)
So given this landscape it makes sense to me why a small team of developers in Red Hat decided to take some of their best work with Ansible, Ruby, Python and APIs to bring IMS into play.
What is IMS?
By now, you might be wondering what this IMS thing is meant to do. Let me quickly sum it up. It’s essentially a toolset built into Cloudforms as of version 4.6 that allows MUCH simpler mass migration of VMs from VMware environments into Red Hat Virtualization environments. This is an ideal way to get a further push to all-encompassing open-source, given that RHV is the supported version of the oVirt hypervisor. oVirt/ RHV is essentially feature-beyond VMware (assuming with VSphere) as of version 4.2, and it has a prettier price-tag too.
1) Have an existing VMware VCenter cluster (or more) with Cloudforms (4.6+) and RHV (4.2+) in a nearby, accessible network.
2) Follow the docs to hook Cloudforms into both VCenter and RHV.
3) Have a planned mapping of datastores and networks to allow for RHV to take-over managing what VCenter did prior.
4) Select the source VMs and set their targets in the RHV environment.
5) Press GO (paraphrasing) and profit! Pending your network and disk performance, all the selected VMs and their resources get moved and ready to run.
- At this point, migrated VMs are in a down-state to allow you to plan for downtime and switching over from the VMware counterparts.
What’s it Made of?
Under the covers, the IMS tool-set uses virt-v2v, with the earlier mentioned combo of Ansible, Ruby and Python. If we compare this tool-set to earlier (un-supported) methods using custom-scripts or hand-bombing one-VM-at-a-time in RHV with VMware as an external provider, it’s obvious that there’s gains that we know well from using automation in general: speed, health, quick repeatability.
As you can see above, the solution can migration into either RHV or RH OpenStack, which is equally exciting if you’ve a private cloud of OSP going on. It’s also worth noting that a best-practice is to have a host dedicated to conversion to keep running clusters un-burdened by the heavy task of virt-v2v conversion.
Meh, is it Worthwhile?
This whole solution might sound drastic. It might sound time-consuming. Yes, it is both of those things and requires planning and careful orchestration of running applications and systems. I argue that this is closely the same effort that goes into migrating applications to container platforms. For those businesses on the path towards more automation, more security or fully Linux-enabled environments, or simply looking for a bridge between closed-source and open-source, this seems like a great option.
Can I See a Demo?
That’s a tall order, given the infra required to show off the tool-set. It is not difficult nor impossible, though. You’ll need at least a single-node of each hypervisor and Cloudforms running on the RHV side. If hardware isn’t easy for you to get, there’s always the option of risking a test with nested-virt. There’s also Packet.net, which offers essentially, hardware-servers-as-a-service, or hardware in the cloud, which could host all of the above for either testing or maybe even as temporary migration nodes as part of the IMS system.
If hardware isn’t a problem for you, there is indeed a Github repo with more details and content about the tool. https://github.com/RedHatDemos/RHS-Infrastructure_Migration
Can Arctiq Help Me With This IMS Thing?
Yes we can. Many cases for using this tool-set come from a larger encompassing project, or the want for a demo or POC environment to see how well it works for business specific workloads migrating from one hypervisor to another. While this may not be a successful pitch, it will at least serve to educate people about something they may not have known before. Personally, I’m happy to show how open-source is destined to become yet even more pervasive than it has become, and any tools that I can use to further that cause are what keep me interested in using open-source tech. Sounds like a win-win to me.
.. and contact us if this is something your organisation might benefit from.