Contain yourselves; Kubernetes will defeat public cloud lock-in

Interview Object storage supplier Cloudian tells us IBM’s $34 billion acquisition of Red Hat  is evidence of a sea change in what ‘cloud’ means. The most effective cloud strategy today typically encompasses private (on-premises and hosted) and public clouds.  This is further confirmation of the hybrid, multi-cloud model, facilitated by Kubernetes. 

How can Cloudian deduce this from the IBM takeover?  Let’s find out Jon Toor, Cloudian’s chief marketing officer, kindly provided the answers.

Jon Toor.

B&F:  Why does the deal speak to a sea change in what cloud means? You say it’s no longer limited to public, remote data centres managed by an external vendor (such as IBM). Instead, the most effective cloud strategy today typically encompasses both private (on-premises and hosted) and public clouds. What aspects of the IBM-Red Hat deal support this conclusion? 

Toor: IT managers expect to employ a diverse array of cloud types, including public, private, hosted-private, and every other permutation you can think of. All of which creates management complexity. With Red Hat, IBM adds the ability to deploy workloads across all these cloud types, whether IBM-based or not. 

Kubernetes orchestration is already widely used, and with this acquisition IBM is adding further weight to a technology that simplifies deployments across on-premises and public cloud computing environments, including AWS, Microsoft Azure, Google Cloud Platform, Alibaba, and of course IBM Cloud.

B&F: Do you see an IBM-Red Hat combination supporting workload movement between multiple public clouds, meaning IBM COS on the one hand and AWS, Azure and GCP on the other? Why?

Toor: Kubernetes is already established as an accepted platform for supporting workload movement between clouds.

B&F: Isn’t having a multi-cloud strategy making IT life harder for enterprises? There are many more compute and storage variations to manage with no one-for-one mobility between the three main providers. Deciding the best value for money requires complex arbitrage. Surely, it’s best to settle on just one cloud supplier?

Toor: Settling on a single cloud supplier may work in the short run. But IT managers know that monogamy is not sustainable for two reasons. First, standardising on one supplier and one set of APIs adds risk and sacrifices bargaining power. Mega-users like Netflix have the power to dictate terms. But most users will be forced to take what’s offered.

Second, there will likely be corner cases where specific users will demand access to other platforms to capitalise on specific feature sets. Cloud suppliers will compete on features and cost and will establish differentiators such as AI capabilities or big data analysis. To some users, these capabilities may be irresistible, thus putting pressure on the one-cloud model.   

Nonetheless, there will still be significant pressure to standardise. This is why analysts predict that in a multi-cloud scenario, 80 per cent of the revenue will accrue to the primary provider.   

B&F: On-premises suppliers would say a multi-cloud strategy including on-premises (private cloud) is the best idea, wouldn’t they? Why don’t they say public cloud is a monopoly supplier disaster waiting to happen – Jeff Bezos is no altruist – and having on-premises IT is the only sure-fire defence against it? Toor: No for-profit company is an altruist. But companies are universally driven by customer needs. On-premises suppliers – and cloud suppliers – exist solely to serve customer needs. And surveys indicate these needs are clear: 75 per cent of IT managers now identify a diverse cloud environment as their preferred platform. Whether they view this as a defensive position or simply the approach that best accommodates their workflows ultimately does not matter.

Decisions are always driven first by the nearest-term needs, and those needs change over time. Walk into any data centre that’s been in service for 20 years and you’ll see a rainbow of vendor colours and names. Cloud suppliers are simply another vendor.     

B&F: If data has a huge gravitational effect and cannot be easily moved then doesn’t that render a multi-cloud strategy worthless? If my data is in AWS and can’t be moved because the egress and networking charges are too high then what does it matter if I have common data management facilities across my AWS, Azure and on-premises IT environments?

Jon Toor: Mobility has value for multiple reasons. First, moving data into the cloud is relatively inexpensive, which can be useful even if the data is never moved out. In disaster recovery planning, for example, data can be replicated to cloud-based cold storage. It will never be touched unless disaster strikes, but there’s great value in the data being safely stored elsewhere. 

Another example is data analysis, where, for example, video content can be loaded to the cloud for image recognition. The resulting metadata can then be downloaded and the cloud-based media deleted. Both use cases employ cloud capabilities without creating un-predictable or excessive charges.

Furthermore, there is great value in mobility even if workload migrations are an infrequent event. For example, a new application could be initially deployed in the cloud to facilitate rapid, low-cost development. As demand grows, the business is free to continually re-evaluate its options and may choose to repatriate that application to an on-premises data centre. 

An IDC survey found that IT managers expect that 50 per cent of applications now deployed in the public cloud will migrate to other cloud types in the next two years. Of course, an even greater number of new applications will likely take their place, starting the cycle over again.  

Comment

It’s a multi-cloud world, with Kubernetes being used to spread workloads across and between clouds. Toor makes a good point when he says a multi-cloud strategy is effectively the only defence against public cloud providers lock in. It’s a customer’s exit strategy negotiating point.

We expect multi-cloud software abstraction layers will emerge to  facilitate single pane of glass management for workload and data deployment and movement and cost reporting and analysis. IT never really gets simpler.

Finally, Toor’s point that public cloud suppliers are just another vendor to add to the existing heterogeneous IT supplier mix resonates strongly with us.

- Advertisement -