Cloud computing is now by no means restricted to enterprise IT. It has brought business value, continuity, and end user utility to organizations of all sizes, across different industries. So much so, that cloud usage is now a given – the Flexera 2020 State of the Cloud Report found that there are hardly any enterprises left that don’t use cloud in any way.
With such widespread adoption, it is imperative that IT teams have no confusion as to the terminology of cloud environments, so that they can choose and implement a cloud management strategy that is right for them.
All cloud environments are built upon one of two building blocks: the private cloud and the public cloud. Private cloud is, more often than not, infrastructure built from utilizing resources within a company’s own on-premises data center. Public cloud, on the other hand, is a set of IT infrastructure and services, including compute, network, and storage resources as well as applications, which are owned and managed by a third-party provider (Amazon AWS, Microsoft Azure, and Google Cloud are the major players) but partitioned and shared between multiple customers.
Pretty much all cloud infrastructure in any organization is built up from a combination of one or more of these private and public cloud deployments, in varying configurations and complexities, according to the needs of their operational workloads.
As complexity increases, hybrid cloud and multicloud models enter the fray.
What is Hybrid Cloud?
Hybrid cloud is an architecture that facilitates the transmission of data and applications between different private and public cloud environments using an encrypted information pathway. A single workload or task that needs to leverage two separate cloud resources – two private clouds, two public clouds, or a mix of both – is made possible by hybrid cloud solutions.
Here are different kinds of scenarios in which the hybrid cloud model comes into play:
- Companies that are managing IT resources using on-premises data centers or private cloud-hosted environments
- Organizations that are migrating from a fully on-premises solution to an environment that uses one or more public clouds in some capacity
- Businesses that are moving back from a public cloud-based environment to a private, on-premises data center
- IT departments that are deploying a Platform-as-a-Service (PaaS) or Infrastructure-as-a-Service (IaaS) solution in which computational resources can be leveraged without significant data risk
No wonder then, that a full 86% of over 3,400 global IT pros surveyed for the Nutanix Enterprise Cloud Index report agreed that hybrid cloud is the “ideal IT operating model” for their organizations. The three most important reasons cited by respondents as a reason for hybrid cloud’s suitability were:
It is clear that cost savings are not as big a driver to the shift to hybrid cloud as other business outcomes. A hybrid cloud infrastructure provides the scale and elasticity of the public cloud while replicating the security of on-premises data centers. It is also simpler for organizations in the healthcare, finance, legal, and telecom verticals to meet compliance and regulatory requirements. The hybrid cloud makes it possible to share data with external partners while conforming to data sharing and storage regulations in their industries.
The Cloud Follows the Workload
About 12% of enterprises across the world currently use a hybrid cloud architecture exclusively for their IT deployments, as per the Enterprise Cloud Index report. That said, enterprises have consistently indicated that the flexibility to choose the best infrastructure for changing workload requirements is paramount to them.
The optimal cloud infrastructure, therefore, keeps changing on the basis of levels of resource usage, costs, compliance requirements, and other business- and technology-influenced factors. A company might want to move workloads between public, private, and hybrid cloud environments, while retaining ease of management or uniform security policies.
The key question for CTOs to ask is what parts would be run and managed by the in-house IT team and what parts would need to be run and managed by a third-party provider or cloud vendor?
While the hybrid cloud makes borders between public and private clouds invisible to users and abstract to IT, what if the workload demands separate private or public cloud infrastructures that don’t need data to be shared between them? This is the domain of multicloud systems.
What is a Multicloud Environment?
Multicloud – the term itself suggests the presence of multiple cloud systems. It is commonly taken to mean the usage of multiple public cloud systems, more often than not from different vendors. The need for such a setup arises from differing requirements and workloads of different departments in the organization.
These public clouds don’t necessarily talk to each other; they just fulfill a specific requirement of a single department or team, which have no business justification for sharing data or apps with another department.
According to the Flexera 2020 State of the Cloud report, 93% of enterprise organizations have a multicloud deployment in place already. They typically realize one or more of the following benefits:
- Businesses avoid vendor lock-in because there are a lot of options to run typical workloads.
- Companies can take advantage of competitive pricing as well as better features and upgrades, again because of the typical nature of the workload.
- The possibility of downtime or data loss due to a localized component failure is minimal.
- Organizations can meet data compliance and regulatory requirements across different geographies by choosing an IaaS provider with physical data centers in specific regions.
What is the Difference Between Hybrid Cloud and Multicloud?
A common misconception is that hybrid cloud and multicloud are mutually exclusive environments or differ much in their characteristics and implementation. In fact, the Flexera report referred to above revealed that 87% of the multicloud implementations involved a hybrid cloud architecture.
Technically, a hybrid cloud is also a multicloud system, as it has at least one private and one public cloud. However, the inverse is not necessarily true, because individual clouds within a multicloud system can exist in silos. That said, a multicloud configuration can be hybridized.
Within both hybrid cloud and multicloud environments, IT admins need to find the right monitoring, analytics, and security management tools that work for them. These might be developed in-house or by a third-party vendor. Whether these tools need to work across different public clouds or not varies on a case by case basis in multicloud deployments.
The key to success with both deployments lies in being able to holistically manage resources as if they were in a single location. Nutanix XI Beam does just that – it goes beyond provider-based optimization tools and delivers complete cloud analytics, centralized cost governance, intelligent consumption planning, and security compliance for complex multicloud environments. Admins can identify and eliminate underutilized resources, compare cloud vendors, and provision the optimal instances for each application with a few clicks.
What all this means is that rather than thinking of hybrid and multicloud as separate solutions for separate scenarios, companies should constantly re-evaluate their suitability and possibility of unification for cost savings and utility for multiple processes and workloads.
Unified Cloud Infrastructure Management
With advances in hybrid and virtualization technology that are increasingly pairing public and legacy systems, organizations can now move to a unified operational model for hybrid and multicloud environments, dubbed the “single pane of glass” by a GigaOm market landscape report.
“Enterprises should invest time and money now into finding the right middleware solution for both hybrid and multicloud,” says David Linthicum of GigaOm.
Growth in cloud-based IT has spawned mature Cloud Management Platforms (CMPs) that wrestle the on-premise, hybrid and multicloud environments into submission. CMPs expedite the ability to connect multiple environments to provide that single operational view. They can provide self-service provisioning, deep analytics, layered and rules-based governance and compliance, faster incident recovery times and more to simplify operations across all business units within an enterprise.
The question is not if an organization should make use of a good CMP, but how to get there? Many enterprises already have the resources needed to take advantage of a CMP but are looking for a definitive map to that destination. There are several considerations and questions to be asked before putting in place cloud management processes:
- Why should an organization pay for a cloud governance tool when public cloud providers have alternatives that are free?
- Are there industry-specific challenges to cloud management?
- How does policy evolve as infrastructure scales?
- What automation opportunities does a hybrid cloud environment offer?
- What should be the criteria for provisioning and de-provisioning with so many cloud apps in the market?
Define and Document an Operational Cloud Strategy
Build and implement a defined, documented cloud management strategy, if there isn’t one already. If there is one in the organization, it’s time to review, update and put it to better use. Here is where the inventory of applications, data, services and residencies are accumulated. The information is collected and refined, allowing the organization to move forward quickly and with confidence in their decisions.
Companies with no time for guesswork know this is critical to providing the path, the size and the scope, pointing everyone in the same direction. Using references from recognized sources minimizes disagreement on meaning and purpose, expediting the discussion.
An effective cloud strategy takes into account facets like:
- Resource usage: Capacity planning can be overwhelming but is very necessary. Optimizing tools built for multicloud environments eliminates underutilized and unused cloud services based on the workload.
- Single vs. Multiple tenancy: Single tenant architectures use software, compute, storage, and networking resources contained in an instance to support the needs of a single customer, offering better security. Multi tenancy utilizes those same resources to support many customers, proving to be extremely cost effective.
- Vendor selection: Consider various cloud providers and players in the market, including managed service providers (MSPs).
- Governance: Automated cloud governance tools deployed in conjunction with best practices help monitor for any anomalies and mitigate issues before they get worse.
Monitor Cloud Usage Costs
Tracking consumption across all cloud resources – per application workload, team, and business units – is necessary to prevent uncontrolled cloud spend and enable accurate resource allocation. This goes beyond contract costs for the current cloud configurations and is more reflection of true TCO. The need is to understand the management and operational costs, as well as the actual man hours required to maintain each of the different environments. This is where a primary business case for a single pane of glass view of all operating environments is found.
This is not a guarantee that all will be strictly cost savings. There are outlays and there will be shifting of costs. The introduction of a single operational model will likely change processes and procedures in multiple areas which will require resources. It may necessitate reassigning existing personnel or acquiring new skills to take full advantage of the product. It may also show that more than one product is required to meet all needs.
Financial considerations must be understood to allow the most prudent decisions to be made. However, they should not be used so rigidly as to limit the ability for the organization to reap the benefits of a single operational model.
Determine Function-Specific Business Requirements
The true TCO is now visible and speaks to the need for a single operational model. With the “what” and the “why” in hand, identify the requirements specific to the enterprise. Prioritizing and documenting the needs, wants and future enhancements allows the conversations and decision to proceed. This can include regulatory, geographic, goal-supporting and financial needs. Every organization is unique, and their CMP should fit their business.
The use cases that every CMP must address are:
- Financial Management and Resource Optimization
- Security, Compliance and Governance
- Provisioning and Lifecycle Management
- Operations and Incident Reporting
- Analytics and Machine Learning
- Third-Party Integration of Apps and Services
An organization’s list will also include other items specific to their configuration. The requirements facilitate the short list of potential providers but should be as complete as possible. Ranking within each category helps to identify the “must have” items should those determinations be required.
Further, all cloud vendors offer extensive support, but some are more proficient than others. Be specific when sharing expectations and how the organization generally works. If someone on-site is required throughout the implementation, make it clear from the beginning. These should be included in the requirements rather than after a provider has been selected.
Mix Cloud-Native Applications with Virtualization
A modern enterprise cloud will have a hyperconverged infrastructure (HCI) that abstracts the underlying compute, storage and networking primitives. This means applications use autonomous “containerized” services managed via agile DevOps processes, with a continuous delivery workflow. Unlike VMs, they scale up and down rapidly, which enables optimization of infrastructure resources.
Andi Mann, chief technology advocate at Splunk, explained the importance of using cloud-native apps. “Taking advantage of cloud services means using agile and scalable components like containers to deliver discrete and reusable features that integrate in well-described ways, even across technology boundaries like multicloud, which allows delivery teams to rapidly iterate using repeatable automation and orchestration.”
The technological considerations for a unified hybrid multicloud environment should be:
- OS-level virtualization: A single OS instance is divided into isolated, dynamically created containers, each with a unique writable file system and resource quota, so that underlying infrastructure dependencies are abstracted.
- Updatability: This is one of the benefits of cloud-native apps, which are always available. In contrast, on-premises apps work on a subscription basis and need downtime when they’re being updated.
- Flexibility: Custom-built applications and services must run on any private or public cloud with little modification so that vendor lock-in is minimized.
- Right-sized capacity: Infrastructure provisioning and configuration is optimized with dynamic allocation of resources. This means application lifecycles and workloads are better managed according to demand.
- Collaboration: The ideal mix of cloud-native and virtualization facilitates improved DevOps, which means people, process, and tools are better utilized in operations to bring application code into production more quickly and efficiently.
Choose the Right Providers
It will be tempting to purchase and install the monitoring product from the largest provider with an existing relationship or the one that seems to have lots of features at a reasonable price. Here is a call to exercise caution. Only by matching the needs of the enterprise to the proposed product will the best fit for the organization be identified.
Pitfalls can include purchasing services that may not be fully utilized, accepting services that appear to cover most of the requirements but may leave some unacceptable gaps, and products that can’t be integrated – especially common with some on-premises, legacy applications. This last case could present a need for a second product or provider. There may not be a perfect answer to every scenario but knowing the exceptions up front is critical to success.
While public clouds today can handle pretty much every workload or scenario where the most stringent data sharing regulations don’t apply, they are optimal only for elastic or “burstable” workloads.
“Predictable workloads can typically be run much less expensively on-premises on world-class HCI,” asserts Steven Kaplan, our VP of Customer Success Finance. “Figure out what the workloads cost to run on each public cloud – whether native or on Nutanix Clusters – versus running them on-prem.”
There are a number of CMPs out there able to integrate on-premises, hybrid and multicloud environments, each with their own services. Finding the one that best suits a given enterprise will depend on the requirements. Being mindful of the goal to simplify the operational model will require looking at several providers very closely.
Having very direct conversations with the short-list of final provider candidates specifically about implementation and product support prior to final selection is advised. Often, this helps make the final decision more clearly visible.
Deploy and Normalize
Once your choices have been made the unavoidable disruption during implementation must be managed. A phased approach is common practice while migrating to newer technologies. The internal team will partner closely with the provider’s team to define timelines and resources required for the migration.
Some environments will move quickly and rather seamlessly. Others may take a bit more work or custom integrations. It is important to in the phased approach to allow learning, adjusting and normalizing to happen along the way.
The post-implementation period will contain growth and present adjustments throughout the organization. Learning the big features and small nuances takes time. Rules are often initially over-tooled. They require review and adjustment, and fine-tuning them without limitations on access and functionality.
Some CMPs include AI and Machine Learning (ML) capabilities in their products. As these are integrated into the organization’s analytics, new patterns and opportunities will emerge.
The whole selection and migration process is ongoing and will take up more time initially, so it is best to start now. Hybrid and multicloud need no longer be synonymous with complexity. It’s time for organizations to claim the simplicity – or single pane of glass – promised by the integrated, converged cloud computing, networking, and storage model.
This is an updated version of the original article that was published on March 10, 2020.
Dipti Parmar is a marketing consultant and contributing writer to Nutanix. She writes columns on major tech and business publications such as IDG’s CIO.com, Adobe’s CMO.com, Entrepreneur Mag, and Inc. Follow her on Twitter @dipTparmar and connect with her on LinkedIn.
© 2021 Nutanix, Inc. All rights reserved. For additional legal information, please go here.