Red Hat Summit: OpenShift As The Next RHEL

Last week I was at Red Hat Summit in San Francisco, and this was my 7th Summit, with three of them as an employee. This year’s summit is by far the largest I have seen in the last several years. One striking part of this year’s summit was how OpenShift is the central theme of Red Hat. As my industry peer Dave Nielsen pointed out, it may as well be considered as an OpenShift Summit than the larger Red hat Summit. This clearly shows the maturity of OpenShift as a platform, its success among the enterprise customers and Red Hat’s realization that enabling application platforms is the next big thing for enterprise IT.

OpenShift as RHEL

OpenShift has progressed from its initial Kubernetes roots to one of the most stable platforms in the Kubernetes ecosystem. With more than 300 customers in the last year alone, OpenShift is turning out to be the next RHEL for Red Hat. When I was the employee of Red Hat three years ago, I was very confident that Application Platforms are the future of enterprises and, with the apparent bias of an employee, firmly believed that OpenShift is going to be the operating system for this application infrastructure. The early releases of OpenShift v3 had its practical issues with complaints regarding deployment and management, but the last few releases have smoothened out most of the quirks. The integration of CoreOS Tectonic helps them make the deployment and management of OpenShift even more seamless. I had open conversations (without Red Hat AR arranging it) with five of the OpenShift customers and everyone one of them echoed this sentiment. In spite of punditry predictions about the doom of OpenShift, they have survived and is on the path to becoming the next RHEL for Red Hat. However, the Kubernetes ecosystem is still crowded, and there are many viable alternatives to OpenShift in the ecosystem. It is important for Red Hat to keep innovating and go beyond the traditional mantra of we make enterprise-ready open source products in the industry.

Highlights

Some highlights of the conference are:

• Red Hat laid out their roadmap for CoreOS integration with their existing products. Tectonic will be integrated with OpenShift to make the deployment, over the air updates and other aspects of operations more automated. The operator framework announced at last Kubecon in Europe will make it easy for ISVs to stand up their application/services on OpenShift. Personally, I believe that the Operator Framework + Open Service Broker API is going to make the Kubernetes ecosystem more vibrant and, possibly, making as the only orchestration layer for enterprises (I will do a post on this topic soon). Red Hat also announced that CoreOS Container OS will absorb parts of Project Atomic and serve as their container operating system. CoreOS Quay registry will provide the container image repository for OpenShift
• The operator framework from CoreOS, announced during Kubecon EU two weeks back, is seen by Red Hat as a critical component to spruce up the partner network around OpenShift, by providing the ISVs a seamless way to automate their deployments
• Microsoft’s announcement, delivered by Scott Guthrie himself that OpenShift will be offered as a service on Azure (and managed by Microsoft) is an exciting announcement. This announcement brings both companies close to each other like never before.  However, it could be an interesting dynamic to watch as Microsoft increases their investments in Azure services based on Kubernetes
• Red Hat has got their story around Cloud Native storage right. They are finally bringing together their past investments on Gluster and Ceph with a unified storage offering around container native architectures. This completes a significant part of their application infrastructure portfolio with Red Hat CloudForms and Ansible filling the other needs
• Red Hat announced OpenShift Cloud Functions, their serverless offering on top of OpenShift. It is based on OpenWhisk and will be initially available on OpenShift Online. It is good to see Red Hat taking initial steps to meet the market needs (See this guest post on serverless adoption being faster than cloud adoption)

What was missing?

Missing from all the conversations around summit is any credible announcement that they will bring Red Hat Mobile investments (they bought a company called Feedhenry) in the context of cloud-native application infrastructure. I didn’t hear much about 3Scale too in the keynotes. With 3Scale, Red Hat is in a perfect position to help the organization manage their Microservices architectures. I am hoping to hear more about these two products in the coming months.

The blind spot for Red Hat

I have been saying this for several years now, asked about it during the analyst day and will emphasize again here. In the modern enterprise era, where clouds dominate the underlying infrastructure, the key to success lies in holding the customer data. This need is the real reason why Google Cloud has invested in their data services and why AWS and, now, Microsoft are doubling up on their data services (see Matt Asay’s article on cloud databases in this context). With Red Hat planning to be a credible player in both Hybrid Cloud and Multi-Cloud world, they need to do both cloud storage and databases right. They have a credible story on cloud-native storage, but they have no story on databases. In the era of cloud computing, the vendor who holds the keys to customer data wins. Red Hat only has half the story right with cloud storage, and they need to invest in getting the cloud database story right. Without a handle to control customers’ data, it is a matter of simple application re-architecture for customers to migrate to public cloud providers. Take the case of Red Hat OpenShift on Azure. With Microsoft exposing their data services wrapped with open service broker API, it is easy to move the applications to, say, OpenShift Online or OpenShift Container Platform on-premises and can be made to work with the data services there. However, all the customer data now lies in Azure CosmosDB or other Azure data services. Data gravity is a big problem, and none of the public cloud providers are going to make it easy to take the data out (in spite of their claims for openness). As an enterprise decision maker who will be facing this scenario, I will invest in making my apps work on Azure Container Service (AKS) than elsewhere with OpenShift. Yes, Red Hat may not feel this pain immediately because most of the enterprise customer data is on-premises, but this scenario will become a pain point much sooner than later. Red Hat should invest in fixing the big hole in their database story.

Disclosure: Red Hat paid for my travel and stay