WavebreakMediaMicro - Fotolia
Two rising stars of the IT world are set to meet as they ascend. The first, hyper-converged infrastructure (HCI), combines servers and storage in nodes that can be built into scale-out clusters. The second, cloud storage, has increased rapidly in use, with hybrid and multi-cloud operations looking ever more realistic options.
So, we have seen suppliers start to combine hyper-converged products with cloud functionality. These cloud-enabled, hyper-converged products allow private cloud operations on-site, while also offering connections to, and interoperability with, public cloud services.
Hyper-converged solutions are well suited to use as a platform for private cloud deployments.
Storage is integrated into HCI products, delivered within a storage controller virtual machine (VM) – in the case of Nutanix, for example – or natively on the underlying hypervisor (for example, VMware with vSAN or Scale Computing with Scribe).
These platforms are application programming interface (API)-driven, with the virtual instance as the central object of management. There is no concept of LUNs or volumes to worry about – the platform manages storage automatically.
Then there is HCI’s easy scalability. Hyper-converged systems are designed to scale by adding nodes or servers to a cluster, with software managing the introduction (and removal) of resources from the cluster, rebalancing as necessary.
And that highlights, in general, why HCI and cloud are such a good fit. The HCI model addresses many of the operational aspects of efficiently deploying and managing physical infrastructure.
So, what should we look for and what is on offer from suppliers in the cloud-enabled hyper-converged market?
Minimum requirements for cloud HCI
First, let’s look at what is needed in cloud-enabled hyper-converged products to provide private cloud services.
- Automation: When resources are purchased in the cloud (such as a VM), the resource is automatically configured for the user. And when that resource is no longer needed, it is decommissioned by the platform. This requires automation at the component level (storage, for example) to create objects required, but also as workflow to compile the right pieces together.
- Marketplace: A marketplace offers services for configuration. In the enterprise, this would be a service catalogue, but in cloud deployments, we can expect to see more complexity. Rather than just offering a VM, the marketplace may provide the option to configure the virtual instance with an operating system (OS), specify memory and storage requirements and deploy an application such as MySQL.
- Multi-tenancy: In public cloud, multi-tenancy equates to separate accounts, but in private cloud, it ensures separate lines of business are isolated from each other from an access (security) and performance perspective.
- Analytics: Increasingly, we see analytics being used to monitor hybrid cloud, to optimise application deployment (choosing the cheapest location) and to locate and retrieve orphaned and unused resources.
If solutions offer some integration with public cloud, then data mobility can be added to the above list. Mobility allows applications or workloads to be moved between on- and off-premise deployments.
Private, hybrid and multi-cloud
As IT organisations look to adopt cloud in various guises, we can characterise progress towards the fullest levels of cloud integration.
Firstly, internal processes often become more cloud-like. Many IT departments will already be well down this route, having introduced service catalogues and billing or chargeback.
Meanwhile, IT organisations (either centrally or through a line of business) often use public cloud to deliver resources for development or as part of a cloud-based strategy.
The challenge then is to align on-premise and public cloud strategies to bring about hybrid cloud. The degree of interoperability achievable between locations depends on a number of factors, including application design, acceptable application latency and the ability to move application workloads between locations.
The final step is multi-cloud, where on-premise operations are combined with public cloud provision from more than one provider. This doesn’t have to mean running all workloads in all locations, but could be as simple as choosing specific providers for particular applications.
Cloud HCI supplier round-up
Dell EMC has a range of hyper-converged offerings under the VxRail brand that cover general-purpose, edge, virtual desktop infrastructure (VDI), performance and storage-focused needs. Based on Dell PowerEdge servers, the appliances use VMware vSAN as the storage layer and vSphere for virtualisation.
Dell EMC now supports Enterprise Hybrid Cloud on VxRail. This is a complete (turnkey) software package that provides the additional features of a self-service portal, integrated data protection and integration with cloud services. These include ServiceNOW for IT service management (service desk functions) and a number of public cloud service providers, including OVH, Amazon Web Services and Microsoft Azure.
Although VMware doesn’t sell hyper-converged systems directly, VMware Cloud Foundation software forms the basis for integrating HCI products with public cloud. Cloud Foundation supports VMware’s Cross-Cloud Architecture strategy, which will integrate with IBM Cloud and VMware on AWS.
Nutanix started as a hyper-converged supplier, but rapidly moved to building out private and hybrid cloud solutions.
Nutanix Enterprise Cloud encompasses a suite of software and hardware solutions that implement the infrastructure components of storage, networking and virtualisation with cloud-focused features of application orchestration and multi-cloud management.
Nutanix acquired calm.io in 2016 and has used the technology to build out Nutanix Calm, an application automation feature. The company also recently acquired Beam, and will use that technology to deliver multi-cloud governance. Finally, Xi provides the capability to orchestrate public cloud and extend into a multi-cloud ecosystem.
HPE offers OneSphere for multi-cloud management. It can be used as software-as-a-service (SaaS) or deployed on-premise, to manage HPE Synergy and HPE SimpliVity converged and hyper-converged platforms.
OneSphere integrates with Microsoft Azure and Amazon Web Services. It can also support VMware vSphere and the KVM hypervisor to support non-HPE platforms.
Cisco Systems acquired CliQr Technologies in 2016 and has used the technology to build Cisco CloudCenter, which provides orchestration and governance for multi-cloud environments. These include on-premise implementations of HyperFlex (with the release of 3.0), Azure Stack, Kubernetes, OpenStack and VMware vSphere. Public cloud support includes AWS, Google Cloud Platform, Microsoft Azure, IBM Cloud and Alibaba Cloud.
Read more on cloud storage
- Computer Weekly looks at the biggest four cloud storage providers, how they stand in the market, the products they offer, and who offers the widest range of products and features.
- We look at what’s possible when it comes to hybrid cloud, what types of data and what hybrid cloud architectures can combine on-premise and cloud storage capacity.
These companies are currently the top four hyper-converged suppliers. Other suppliers are orchestrating public and private clouds with solutions that support hyper-converged and virtualised environments.
Red Hat offers CloudForms, which is based on the open source ManageIQ project and Ansible automation tool. CloudForms works with existing orchestration tools such as OpenStack Orchestration, AWS CloudFormation and VMware vApp to implement an application service catalogue.
ZeroStack enables the deployment of on-premise infrastructure running managed OpenStack, with management through the public cloud. ZeroStack’s Intelligent Cloud Platform is a SaaS offering that also integrates with AWS and Azure in the public cloud and vSphere and Red Hat in private cloud deployments.
Morpheus Data provides support for a wide range of public cloud and on-premises solutions. The Morpheus system enables automation and governance features, while extending capabilities to include cloud brokerage. IT organisations can compare costs of deployment and place applications in the most appropriate location by price and features.
Finally, it is worth pointing out that although this article covers hyper-converged, private cloud does not necessarily need HCI. There are solutions from Cloudistics and Stratoscale, for example, that provide private cloud capabilities without HCI.
These platforms implement hyper-converged scalability features within their own software and are a worthwhile alternative when HCI is not an option. This could include the requirement to use existing hardware and licences or hardware platforms not supported by the HCI suppliers.