Cisco and Google have announced a partnership aimed at delivering hybrid-cloud solutions for enterprise customers, just one of the more notable of a series of developments that highlight how important bridging the divide between on-premises systems and public clouds has now become to corporate IT strategy.
Cloud services as we now understand them have been around for at least a decade, with Amazon’s launch of its Elastic Compute Cloud (EC2) and S3 object storage offerings. Adoption has increased gradually since then, as organizations figured out how to make best use of cloud services and where these fit into their overall IT strategy.
Initially, this consisted of using the cloud for development work and deployment of public-facing applications and services with other enterprise use cases such as backup and disaster recovery following. Some organizations have also experimented with shifting workloads to the cloud, with varying degrees of success.
“What’s happened is that the move to the cloud started off as being pretty chaotic, and it was pretty much business-productivity type of workloads that were the first to get moved there,” said Roy Illsley, lead analyst for cloud and virtualization at Ovum.
Now, organizations are looking at moving core business workloads into the cloud, but they are finding that just taking multitier applications and dumping them onto a public cloud in a so-called lift-and-shift operation does not always give satisfactory results.
The trend now emerging, according to Illsley, is a consensus that migrating core business systems to the cloud will involve a multistage process. Building a private cloud can be seen as the first step, so that core workloads can be moved there in order to gain understanding of how things operate in a cloud environment before proceeding further.
Proceeding further may involve refactoring or rebuilding applications to make them cloud-native, or at least better suited to operating in a cloud environment by being infrastructure-aware and able to scale themselves up and down as required.
However, once you have resources both on premises and on a public cloud, you are already running a hybrid-cloud environment (or even multi-cloud, if you use more than one cloud provider). And what many in the industry have come to realize is that some measure of consistency is required between the public and on-premise sides of the equation, in order to make a hybrid cloud operate more smoothly.
Vendors partner up
This is what many recent hybrid cloud developments appear to be striving for. The Google and Cisco partnership, for example, focuses on a container-based strategy using the Kubernetes orchestration tool to help customers develop and deploy cloud-native applications and services both on-premises and in Google’s cloud.
Kubernetes was developed by Google and is already used in its cloud to drive the Google Container Engine service. What the new partnership brings is Kubernetes support on Cisco’s HyperFlex hyperconverged systems, along with another open-source tool, Istio, that provides cloud-like load balancing and automation capabilities.
The underlying HyperFlex platform is Cisco’s flavor of hyperconverged infrastructure, which copies the way hyperscale providers such as Google build their own infrastructure – using clusters of server nodes that operate a software defined storage pool from the aggregated direct attached storage resources across the cluster.
IBM has also followed the Kubernetes path with its recently announced IBM Cloud Private software, which the firm explicitly states is “designed to enable companies to create on-premises cloud capabilities similar to public clouds to accelerate app development.”
This is based around Kubernetes, Docker containers and Cloud Foundry, and is designed to make it easy to connect with workloads running on IBM’s public cloud or even migrate them there. It can run atop a variety of infrastructure platforms, including IBM’s Z mainframes and hyperconverged systems based on Nutanix software.
Where IBM’s offering stands out, however, is that it has also released container-optimized versions of key enterprise applications and services, such as the Db2 database management system, MQ messaging middleware and WebSphere Liberty.
Other major players in the IT market have, like Cisco, focused on hyperconverged systems to simplify on-premises infrastructure and make it a closer match for that operated by the cloud providers as part of their hybrid play.
Microsoft has introduced its Azure Stack, for example, which aims to let customers have an on-premises version of its Azure public cloud, while VMware has introduced a platform that can be deployed either onto on-premises server hardware or in a public cloud.
Azure Stack is designed from the outset as a hybrid-cloud platform, although one that is tied to Microsoft’s own Azure cloud. It is designed so that organizations can operate Azure services from their own data center, and manage them using the same management console used to oversee Azure services online.
The platform is only available as a pre-integrated, hyperconverged offering from key hardware partners, which currently comprise HPE, Lenovo, Cisco and Dell EMC. Microsoft claims that this is to ensure that an Azure Stack deployment performs optimally, but it means that buyers have to invest in new infrastructure to get it.
Meanwhile, VMware introduced its own infrastructure platform called Cloud Foundation, which includes its vSphere, vSAN and NSX products in a complete software-defined infrastructure stack. This can be deployed onto compatible hardware, which includes hyperconverged systems hardware from VMware’s parent Dell EMC, Hitachi Data Systems and Fujitsu, among others.
However, Cloud Foundation is also available from several major cloud providers, including IBM Cloud, Rackspace, CenturyLink, and as a special VMware Cloud on AWS service that is managed and supported by VMware itself. This results in what is effectively a private-cloud environment, but one that is hosted on the public cloud provider’s infrastructure.
Both Microsoft and VMware’s solutions thus revolve around having a consistent platform at both the on-premises and public cloud ends of the connection, but the two firms have different aims.
Microsoft largely sees Azure Stack as an on-ramp to its Azure public cloud, enabling customers to develop and test workloads internally before deploying them on Azure. However, the firm also concedes that some workloads will need to be kept on-premises for customers in highly regulated industries, and Azure Stack allows for a hybrid application where sensitive data is processed locally while the rest of the application runs from the public cloud.
VMware, in contrast, seems to view Cloud Foundation as a way of staying relevant to its traditional enterprise-customer base as they adopt a cloud strategy. The platform enables those firms to move some workloads to the cloud, potentially saving on not having to provision physical servers, while staying with the platform and tools they are familiar with.
For VMware’s cloud-provider partners, there is the opportunity to pick up some of that cloud spend, while Amazon is quick to promote the fact that VMware Cloud on AWS provides seamless access to its native AWS cloud services for future development.
Moving forwards
The current trend then is that enterprise IT vendors have realized they need a best-in-breed cloud presence to sell to their customers, while the cloud providers have realized they need a best-of-breed on-premises partner to deliver a bridge from the customer’s data center to their platform.
“What we are seeing now is the emergence of these large trading blocs. You’ve got AWS-VMware, you’ve got Google-Cisco, you’ve got OpenStack, you’ve got Microsoft, and they’re each pushing their own on-premises-to-cloud route as the most direct, but they do all recognize, even if reluctantly in the case of AWS, that organizations are being strategic and are now using multiple clouds defined by the characteristics they want from those clouds,” said Illsley.
The next challenge for organizations is how to make the new workloads that they move into the cloud operate wherever they want them to, while maintaining a level of visibility that will allow them to see whether they are getting the best value or not.
This means that cross-platform monitoring, metering and management tools are required to provide IT administrators with the ability to oversee a range of workloads that may be operating on-premises and on more than one public-cloud platform.
This part of the market is still at an immature stage. The incumbent vendors tend to focus on tightly integrating their own tools with their own platform and maybe one or two cloud providers, while third-party tools may support multiple platforms but their depth of support tends not to be as great because of that.
VMware is one firm working towards this with its recently announced VMware Cloud Services include tools to manage services running on VMware-based infrastructure as well as public cloud services such as AWS and Microsoft Azure, and compare and analyze the costs associated with each one.
Ultimately, the winners of the hybrid-cloud race may be those that can deliver a genuine cross-platform capability that lets organizations pick and choose the best place to run their workloads. The big players see this as an opportunity to keep the enterprise IT spend flowing their way rather than into the pockets of cloud-first startups.
“If you want to attract people to your platform they are not going to come just because you’re sexy and new, it’s because you’ve got a path that will get them to the cloud,” said Illsley.
This story, "What does hybrid cloud mean in practice?" was originally published by IDG Connect.