IBM bought Redhat this week; predictably, the HN thread is full of woe.
RHEL is the obvious core concern here, but Red Hat are also caretakers for a whole lot of other open source awesome you almost definitely use (or have heard of, or benefit from):
- Fedora / CentOS (upstreams of RHEL)
- Ansible (Python+SSH based configuration management)
- OpenShift / Quay (commercially supported K8s distribution / container repository)
- CephFS / GlusterFS (cloud native clustered storage)
- Active participants in Linux kernel development
Given the land rush still occurring in this space and the instability resulting, one can make the argument that - for consumers - there is inherent value add to vendors and solutions that are either pure open source - not reliant on a commercial entity - or those championed by corporate entities not at risk of acquisition, for example due to their own prestigious market cap (FAANG et al).
Who next? #
First that comes to mind is Rancher Labs. Everything they do is great or on the cusp of being so, yet they’re small and ahead of the curve which ultimately means affordable. They’re ripe for acquisition. By who, and what happens afterwards, are a known unknown. Building your mission critical [compute, orchestration, distributed storage etc] on a commercially supported platform with an uncertain future - despite its present day potency - is risky. Risk is exactly what enterprises pay through the nose to avoid.
The same argument can be made for Hashicorp (Terraform, Vault, Vagrant, Consul, Nomad). These tools are a mainstay in the DevOps space, and for good reason. Hashi is a small fish, though apparently quite profitable (thanks to lucrative enterprise support contracts) - in a big pond, which makes them [expensive] prey to deep-pocketed juggernaut; is your companies whole infrastructure [as code] strategy reliant on a product that can could be mothballed tomorrow by an archaic behemoth (hi, Oracle/IBM/CA)?
One step further: VMware’s legacy business is on the way out; the hypervisor has been a commodity for a half decade now. They’re thus an ever more tempting sell-off asset for Dell EMC, so unless their in-progress pivot to container relevance really takes off, adopting even their ‘cloud native’ offerings involves risk.
Nov 7 Edit: VMware has just acquired Heptio, emphasising this trend. Heptio are leaders in the managed K8s space, offering enterprise support so long as you deployed their reference architecture, and relatively widely adopted ecosystem additions like Ark (subsequently renamed to Velero).
Recommendations? #
DIY everything is less insane than those who’s confirmation bias about their all-in-on-AWS strategy would have you believe. Sure, you’ll pay more to attract and retain high calibre headcount, but then get that back (and more) in the huge opex reduction that comes from commodity hardware + open source software, instead of paying a cloud vendor to sell you a boilerplate version of the same, as a service (, with a moat). Failing that:
Azure / AWS / GCP aren’t going anywhere.
Neither is NetApp (who will gladly sell you a HCI centric bare metal variant of K8s).
Bottom line, make sure your vendor of choice has a roadmap - and market trajectory - aligned with yours. For example, many look forward to GOOG’s teased GKE-on-prem offering (because GKE on GCP is best in class), but what incentive does one of the worlds biggest cloud companies have to make their flagship cloud product work exceptionally well on your infrastructure? GKE in and of itself has lured a deluge of loyal AWS customers onto GCP because it’s that much better than Amazon’s offering. There, those loyal AWS customers have accidentally found a competitively priced, performant competitor to a vendor they had previously considered peerless. These ‘serendipitous’ discoveries would not have occurred without the carrot of GKE, so why would GOOG make self hosting GKE-style K8s equally attractive to the on-GCP variant?
Similarly, MSFT, while historically an ally of the jack-of-all-trades SME sysadmin, is now all in on their own cloud offerings, as evinced not only by their huge and continuing investment in Azure, but also their very intentional messaging at Ignite this year. Case in point: a telegraphed phase out of their widely adopted Skype for Business (nee Lync) on-prem offering, in favour of O365-only (SaaS) Microsoft Teams - a great product, but a marked departure from their previous ethos of proudly allowing customers to choose where to deploy their workloads. If you want to use Microsoft’s [excellent] UC platform post 2023, you’re doing it on their terms: by subscription, in their walled garden.
Caveat emptor.