The Future Is Hybrid

The journey from data centers to the public cloud started more than a decade ago, allowing small startups and medium-sized enterprises (SMEs) to scale fast with relatively low costs.

The overall cloud market future seems to be very promising. In 2021, the overall cloud market share was USD 494.7 billion. By 2026 is expected to almost double to USD 947.3 billion,

According to Gartner, by 2025, 80% of businesses will try to move their computing to the cloud. In addition, Gartner predicts hybrid cloud Market will triple by 2026. The hybrid cloud was valued at USD 52.16 billion in 2020 and is expected to reach USD 145 billion by 2026.

The hybrid cloud market is anticipated to emerge as the fastest-growing segment owing to the growth of cloud and industrialized services and a decrease in traditional Data Center Outsourcing (DCO), as suggested by Grand View Research.

In this article, I will try to explain what is hybrid cloud.

Why is it growing so fast? And why I believe the future of the cloud is going Hybrid.

What is a hybrid cloud?

A hybrid cloud setup combines a public cloud with a private cloud or on-premises infrastructure. On-premises infrastructure can be an internal data center or any other IT infrastructure that runs within a corporate network.

In other words, it allows companies to expand existing infrastructure using the cloud.

The image below illustrates how the cloud has evolved.

Originally cloud was offered as a solution to extend or outsource enterprise-owned data centers. As a result, more technologies to improve cloud infrastructure were introduced by major vendors.

Public clouds have evolved dramatically, offering highly secured and better performance at a relatively cheaper cost. It was a very appealing offer for small companies that couldn’t afford to build their own data center. But still not enough discount for big companies. Those companies preferred to invest in a private cloud, running the cloud technology on their data centers.

That leads us to our point today, where organizations have many options to pick from.

Hybrid Cloud Common Use Cases

Cloud bursting — Handling expected and unexpected loads.

Imagine a country that hosts the Olympic games. For a limited time, Telco service providers might have massive expected loads. A decade ago, the telecom company would have had to purchase additional hardware to meet those peak loads, but after the Olympics, that additional hardware would no longer be needed. By applying the cloud bursting configuration, service providers could overcome the overload simply by spinning up additional loads on a public cloud and distributing the loads across.

The spin-up process can be triggered automatically when a request load is reached or scheduled manually for a limited time.

See this link for a multi-cloud bursting demo using open-source tools.

Cloud specific Services — Cloud provider used for complementary services

Specific services provided by the cloud are also common use cases. Good examples could be image recognition (e.g., Google Cloud Vision), speech-to-text (e.g., Amazon Transcribe), Azure Machine Learning, analytics processing, and more.

For example, Machine learning requires dedicated hardware like special GPU chips for better performance. Usually, this hardware is expensive and does not always require constant resource amounts. Using a “Unique services” pattern where regular workloads are on-premise and ML training or Analytics are running on the public cloud for a limited time can dramatically reduce costs.

Disaster Recovery (DR) — Always ready for an unexpected system failure

A disaster recovery site requires at least one remote (different geolocation) data center with additional hardware, IT admins, and security.

This pattern includes an active-active design with a primary “on-prem” site and “in-sync” cloud DR. Alternatively applying an active-passive cloud DR where the persistent components are frequently updated and backed up. The passive DR will be fully available only in case of a disaster.

Building a cloud DR site in case of failover has considerable benefits.

Multi-geography DR site is an easy task when it comes to the cloud setup. Additionally, most cloud providers are massively investing in storage and backup technologies to manage massive data storage efficiently; therefore, they can offer a variety of reliable cold or warm backup solutions.

The overall cost of maintaining a cloud DR site is reduced dramatically.

However, keeping the cloud DR site in sync is complex and requires analysis and planning.

You can read here more about implementing hybrid DR implementation using OpenShift.

For more patterns, I recommend reading Gregor Hope’s Blog: Hybrid Cloud: Slicing the Elephant

Hybrid cloud challenges

Hybrid cloud is still in its early stages and needs to overcome critical issues before being fully adopted. I have listed some problems I have dealt with recently.

Portable application

Transitioning to the hybrid cloud requires an in-depth review of the existing software architecture and applying suitable patterns.In most cases that I encountered, existing applications are unsuitable for hybrid cloud. Some examples could be:

• The old monolith application operates well on VM. Still, it relies on specific OS features or network resources that are being used differently or not available when moving to the public cloud. A possible solution could be running this application on a containerized VM like kubevirt
• Workloads that have access to a service or API provided by a specific cloud provider. When moving to a hybrid cloud, those cloud services should be accessible no matter where the container is located.
• Container images with “backed-in” configurations that require a specific infrastructure (strictly only on x86 processor arch etc.)
• Stateful workloads where states are stored in a limited access storage solution
• Containers built on one hardware architecture may not fit into other hardware architecture; therefore, they will fail to run.

Managing multiple infrastructures

Hybrid cloud architecture requires mastering different infrastructures, OS, virtualization, and containers. For example, an organization might have Linux servers, Windows servers, VMware, cloud workloads, and self-managed orchestration tools. Ensuring security patches are applied on all server types leads to a complex IT process and administration, resulting in higher costs.

Cross-Cloud networking

Establishing a cross-cloud secure and fast networking is another challenge. Ideally, a service on one cloud should be able to communicate with another service easily.

Distributed workloads

Managing Distributed workloads will require a robust orchestration tool. Similar to Kubernetes but for hybrid workloads. This orchestration tool should be able to ensure security policies and permissions across all clouds. Deploy workloads using rule engine and CI/CD pipelines. Collecting stats from different clusters for Monitoring and alerts. Establishing networking connectivity across clouds

All big players understand that a one-stop shop orchestration tool is a crucial component for leading the cloud market. Google is working on Anthos. AWS has Outpost, Azure arc. Red Hat offers a Kubernetes-based approach offering Advanced Cluster Management or the open source version Open Cluster Management.

Although those platforms are slightly different in their offerings, they all try to solve the hybrid-cloud orchestration issues.

Hybrid cloud open source projects

After listing hybrid cloud challenges, I realized it wouldn’t be appropriate if I won’t propose a few emerging open-source projects I am familiar with that will overcome those challenges and help build a reliable open hybrid cloud.

Note that all those open-source projects are Kubernetes add-ons. Kubernetes has been widely used and offered by all the public clouds for many years, making it a reliable, common ground platform for running hybrid cloud workloads.

Orchestration

• Open Cluster Management — cloud vendor-neutral platform for deploying workloads across multi Kubernetes clusters. Open Cluster Management offers a flexible and rich rule engine for placing workloads transparently. The Open Cluster Management (OCM) architecture uses a hub-agent model. The hub centralizes control of all the managed clusters. An agent named as klusterlet, resides on each managed cluster to manage registration to the hub and run instructions from the hub.

Network

• Skupper — Multi-cloud Kubernetes cluster addon that established layer 7 service interconnect. Enables secure communication across Kubernetes clusters.
• Submariner — Hybrid/multi-cloud Kubernetes addon that established cross-cluster L4 connectivity using between pods and services

Security

• Stackrox — This project is not only hybrid cloud-oriented. However, it’s very helpful, especially when running workloads on multiple environments. The StackRox Kubernetes Security Platform performs a risk analysis of the container environment, delivers visibility and runtime alerts, and provides recommendations to proactively improve security by hardening the environment.

Automation

• Argocd — Argo CD is a declarative, GitOps continuous delivery tool for Kubernetes. a hybrid cloud deployment can be achieved by combining with OCM

Storage

• NooBaa — Hybrid cloud storage solution — NooBaa simplifies data administration by connecting to any of the storage silos from private or public clouds and providing a single scalable data service using the same S3 API and management tools. NooBaa allows full control over data placement, letting you place data based on security, strategy and cost considerations in the granularity of an application.
• Rook — Rook is a cloud-native storage orchestrator that turns storage software like Noobaa into self-managing, self-scaling, and self-healing storage services. It automates deployment, bootstrapping, configuration, provisioning, scaling, upgrading, migration, disaster recovery, monitoring, and resource management. Rook uses the facilities provided by the underlying cloud-native container management, scheduling, and orchestration platform to perform its duties.

Others

Knative — provides an enterprise-grade serverless platform that brings portability and consistency across hybrid and multi-cloud environments. Knative offers features like scale-to-zero, autoscaling, in-cluster builds, and eventing framework for cloud-native applications on Kubernetes. Whether on-premises, in the cloud, or in a third-party data center, Knative codifies the best practices shared by successful real-world Kubernetes-based frameworks.

Hybrid is the reality.

Having listed the problem above, I believe that hybrid-cloud is the next evolution of the cloud market simply because it’s a market need.

A significant portion of enterprises already consumes from more than one cloud. In addition, enterprises are looking for agility and won’t accept vendor locking. Therefore, hybrid or multi-cloud solutions have become a core concern for Operation.

In addition, edge computing workloads have also started to gain traction. The need for hybrid-cloud, multi-cloud, and edge creates a strong motivation for stakeholders such as cloud providers, software vendors, and open source communities to offer proper solutions.

Secondly, in reality, many companies can’t just move to the public cloud.

“No CIO will wake up one morning to find all his or her workloads in the cloud. Hybrid cloud is a reality.”

Over the years, many enterprises have invested heavily in their data centers and can’t just abandon the hardware and move to the cloud.

In addition, because of existing regulations prevent moving sensitive data to the public cloud, especially in the Military, Finance, and Insurance industries.

Those companies are looking for hybrid cloud solutions to take advantage of both private and public.

Conway’s law suggests, “Any organization that designs a system (defined broadly) will produce a design whose structure is a copy of the organization’s communication structure.”

The relevance of Conway’s law is mind-blowing when I think about it in the context of hybrid cloud future adoption.

Since the pandemic, the Tech industry, in particular, has fully adopted the hybrid work model.

On any given day, a team member can choose to perform his tasks from the office or a remote location. Following Conway,

Shouldn’t we expect a reflection of hybrid work in future hybrid-cloud systems?