The applications of tomorrow
Containers are everywhere.While containerization is great for scale and agility for any kind of app, not every app needs to be broken down into micro-services. In my previous blog post on Containers + VMs > IaaS or PaaS, I touched upon the benefits of monolithic containerization from a cost perspective. Monolithic or micro-services, containerization is, however, the underlying theme, for the majority of the apps of the future. If this is going to be the case, shouldn’t the infrastructure of the future be ready for these containerized apps of the future?
Why is the current infrastructure not suitable for the applications of tomorrow?
Let’s look at both the hardware & software aspects of the infrastructure.
- Hardware –
The growing success of hyperconverged players like Nutanix and the increasing push from incumbents with traditional/converged hardware stacks to now offer hyperconverged appliances, is a clear indication that enterprises are rapidly adopting hyperconverged (or hyperscale depending on how detailed you want to get here) as the winning hardware approach. The primary reasons seem to be driven by capex, opex & management simplicity. This market is projected to cross $10B in the next 5 years.
- Software –
The primary underpinning software in the data center or the cloud, for the last decade and a half, has been the hypervisor. While that is great for traditional applications or apps where containerization doesn’t offer benefits, as we move along in the container adoption cycle, it is getting very clear that a ‘Container-as-a-Service’ platform, analogous to what the hypervisor has been in the VM world, needs to be available.
While it is certainly possible that a CaaS platform can be setup on top of the hypervisor, if you can avoid the hypervisor layer altogether and save time/cost/effort, why wouldn’t you!
In reality, based on the learnings the last year and a half, most enterprises would want the flexibility to run containers on both VMs and bare-metal (and the cloud).
Hypervisors + Hyperconverged infrastructure is not going to be a complete stack when containerized apps enter mainstream enterprise line of business work-loads. While it certainly has been functioning pretty well for traditional apps so far, as the underlying CaaS platforms mature, the infrastructure would need to support not just VMs, but containers running natively on bare-metal (and other substrates like VMs on-prem and VMs on the cloud)
One could argue that the security posture of a CaaS running directly on hardware isn’t yet ready for primetime, especially in the CISO’s eyes. There is certainly truth to that, and it also explains why most of the container efforts and POCs today are occurring on VMs. Once these efforts go beyond POCs, and once the security efforts from both incumbents as well as new entrants mature to address the ‘CISO peace of mind’ issue, we will start seeing more and more containers run on bare-metal OSes.
Meet Hyperconverged Container-as-a-Service platforms
The ideal infrastructure of the future datacenter, is therefore, a hyperconverged appliance that can not only run a hypervisor, but also containers natively on bare-metal. Having said that, the flexibility to expand into VMs on-prem or the cloud is key, and that is where CaaS platforms will strive to go.
If you are a hyperconverged appliance vendor, chances are you are already thinking about your container strategy.
Now is a perfect time to get started and think about offering a hyperconverged appliance that comes with a built-in CaaS platform, so you don’t miss the boat when containers become a first class citizen!
If you’d like to work with ContainerX to offer such an all-inclusive solution to your customers, we’d love to chat.
ContainerX is a silicon valley company founded by PhDs from VMware, Microsoft, Citrix, building a ready-to-go container infrastructure platform designed for enterprise IT. It enables enterprise IT architects to setup an elastic multi-tenant & resilient Container-as-a-Service platform for both Linux & Windows environment, allowing developers and Ops to self-service.