momius - Fotolia
A few years back, VMware (and the like) were fond of proclaiming that the only cloud worth considering was a private one, with suppliers and users alike often citing performance and security issues as reasons not to go to the public cloud.
Despite this, the public cloud offerings of Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform have proved popular with enterprises, while various private cloud technologies have withered on the vine.
The problem private cloud has is that many suppliers decided against finding a standardised platform that could interoperate with other private and public clouds.
HP (with its original Helion platform), Dell, IBM and others all offered their own flavours of cloud – to a deafening silence from purchasers in their target markets.
Most organisations do not have enough suitable applications to make private cloud work properly either, whereas a public cloud platform can run anything from tens of thousands to millions of workloads across a shared resource environment. The average private organisation may have up to a few hundred.
Another factor is that so few applications are cloud-optimised: just taking an existing SAP or Oracle application and bunging it on a cloud-capable platform does not make it a bone fide cloud application.
Quocirca sees quite a few “cloud” implementations of this kind, yet – when the details are investigated – all it amounts to is some advanced virtualisation, with no sharing of resources across multiple applications.
A private cloud may be running at only 40% resource utilisation, yet could still be hitting resource ceilings when demand gets high, while a public cloud should be running at above 80% of resource utilisation.
The second coming of private cloud
Things are slowly changing, with many suppliers throwing their weight behind OpenStack and providing a much greater level of fidelity across hybrid clouds, particularly where other de facto standards for public clouds (such as S3 storage compliance for AWS) are brought in.
The adoption of DevOps is beginning to move organisations away from the monolithic application approach to a broader microservices one based around container technologies, such as Docker.
This is driving a need for a highly dynamic platform where containers can be provisioned rapidly against resource pools – just what a cloud is designed for.
The fact that organisations want to choose where a workload resides also drives this continuous delivery (CD) approach towards a cloud-based platform.
So, what does this mean to an organisation looking at implementing a cloud platform?
- Understand what power a true cloud platform brings to the organisation. It is not just advanced virtualisation; it is not just a means of hosting existing applications. Embrace the concept of running multiple different workloads against a pool of shared resources.
- Identify applications that could be moved to a cloud platform now. They may be new applications, existing ones or software being developed in-house using DevOps.
- Do not assume all workloads belong in the cloud: it may be that services called by a monolithic application are ideal for placing in the cloud, whereas the application itself is not.
- Figure out whether, and at what speed, your organisation plans to adopt a full hybrid cloud environment, and approximately what percentage of workloads are expected to be running on the public cloud component after a given period of time.
- Decide what you want as “private cloud”. A private cloud is not always predicated on it being hosted within your own datacentre: it can be co-located or a private instantiation of a cloud managed within a third party’s environment.
- Ensure the adopted private cloud system supports all the standards required to both carry the expected workloads and to interoperate with the public clouds expected to be in use. If Azure is going to be your main public cloud platform, then there will be greater standardisation by choosing Microsoft Azure Stack as your private cloud platform.
- Look for cloud management software that can provide the capabilities around provisioning, monitoring, remediating, updating, moving and end-of-lifing workloads across the resulting hybrid cloud.
- Be concerned – but not paranoid – about security. Securing your organisation’s environment is likely to be far more troublesome than protecting a public cloud platform, all of which operate on a shared security model.
- Ensure the chosen private cloud platform is inherently secure, and that updates are applied regularly. Where possible, use third-party, cloud-based security systems to monitor and manage the private cloud, because these will often be more up to date and effective than in-house systems.
In some cases, a “private cloud in a box”, where an appliance approach is used to simplify private cloud provisioning, may be worth considering, such as is offered via Nutanix, Nuvlabox or ZeroStack.
Such a basic checklist should ensure a private cloud platform can be chosen that will not result in the problems encountered by those who went down the proprietary private cloud platforms a few years ago. The chances of having to reverse out of any decision will be minimised.
Read more about private cloud
- Enterprises have come to view private cloud as a “toxic” phrase, but new ways and means of building deployments is changing that, according to the OpenStack Foundation and its stakeholders.
- OpenStack costs less to deploy than Microsoft, VMware or Red Hat private cloud, but engineering shortages drive up its total cost of ownership.
As a final note, it is worth mentioning that, having got private cloud wrong in the past, HPE, VMware, Dell, IBM and other incumbent suppliers now have offerings that are far more standardised and suitable as a private portion of a hybrid cloud platform.
The “Wild West” early days of private cloud are now in the past. Progress has resulted in a far more standardised and usable approach to how private cloud can offer distinct value in how an organisation operates its overall IT platform.