OpenShift is Red Hat's Cloud Computing Platform as a Service (PaaS) offering. OpenShift is an application platform in the cloud where application developers and teams can build, test, deploy, and run their applications.
Openshift provides a way to deploy Containerized Java applications running on JBoss EAP and RHEL in a very easy way. Reducing server preparation time, hardening-server headaches, and deploying to different stages nightmares.
OpenShift is a CAAS (container as a service platform). This is a virtualization technology that can compare with traditionnal VAAS (VM as a service) technologies but Openshift is lightweight and fast. Everybody in the IT world is now going to Containers. This is a disruptive …
I really recommend the platform because it runs in any kind of environment (cloud, on-prem) and can be migrated between different clouds with no big impact on the company. The possibility of having exactly the same platform across different clouds is a huge potential that puts Openshift ahead of other solutions.
JBoss EAP is subscription based/open source platform. It's very reliable and great for deploying high transaction Java based enterprise applications. It integrates well with third party components like mod_cluster and supports popular Java EE web-based frameworks such as Spring, Angular JS, jQuery Mobile, and Google Web Toolkit.
Obviously, it does container orchestration very well because it's the main purpose of the product. Creating routes has been very easy with the product, creating accessible routes, managing quotas, managing our workspace and our workload has been very efficient with this as well as managing our horizontal scalability.
MOD_CLUSTER integration. JBoss EAP integrates pretty well with mod_cluster. This is an intelligent load balancer especially useful in highly clustered environments.
Supports enterprise-grade features such as high availability clustering, distributed caching, messaging etc.
Supports deployment in on-premise, virtual and hybrid cloud environments.
Upgrades can be stressful to watch - often cluster operators report that they are failing even though the upgrade is proceeding without issue.
Native observability is lacking. The canned dashboards are great, but to really dive into an issue, you need to be proficient in PromQL or have a third-party product installed for correlation.
It would be nice to be able to review logs for containers that have been removed, at least for a few hours after they're gone. Instead we have to rely on log aggregators to view historic details.
Jboss CLI is a great tool but we had trouble using it to get values that are displayed on Jboss GUI. It also has limitations parsing the applications.xml files and we had to use a mix of jboss-cli and linux bash commands to automate certain application administrative tasks.
JBoss doesn't really provides performance tuning recommendations. It would have been nice if it could learn from the current demand vs current settings for things like connection pool, server configurations, garbage collection etc.
Leverage OpenShift Online constantly at both the free and paid tiers. While AWS is convenient, it often brings more administration than I want to deal with for a quick application (i.e. Drupal or Wordpress blog). OpenShift also simplifies the DNS registration and ability to share application environments with team members
As I said before, the obserability is one of the weakest point of OpenShift and that has a lot to do with usability. The Kibana console is not fully integrated with OpenShift console and you have to switch from tab to tab to use it. Same with Prometheus, Jaeger and Grafan, it's a "simple" integration but if you want to do complex queries or dashboards you have to go to the specific console
Applications deployed to OpenShift clusters stay responsive when peak load hits or when the traffic dies down - since the platform reacts by scaling out or scaling in the deployed applications elastically - achieved through' policy sense and response automation - leveraging monitoring, measuring (metrics), auto-scaling to meet SLAs, SLOs, and SLIs. This approach works for stateless or stateful business logic hosting applications. The deployed applications perform consistently, stably, and securely across many deployment platforms - public clouds, private data centers, at the edge, or on factory floors - hosted by bare metal or virtual environments.
Usually, Red Hat JBoss Enterprise Application Platform is good at performance and well suited for high traffic Java EE-based applications, but we have faced hard times performance tuning it for our specific needs. The product would be nicer if they would add a performance diagnostic and recommendations feature to it.
Their customer support team is good and quick to respond. On a couple of occassions, they have helped us in solving some issues which we were finding a tad difficult to comprehend. On a rare occasion, the response was a bit slow but maybe it was because of the festival season. Overall a good experience on this front.
From what I remember of Pivotal Cloud Foundry/Tanzu, OpenShift has a lot more features. Ultimately, our team went with OpenShift. After trialing multiple providers, we decided OpenShift was the best experience. We also had more confidence in Red Hat.
It's easy to understand what are being billed and what's included in each type of subscription. Same with the support (Std or Premium) you know exactly what to expect when you need to use it. The "core" unit approach on the subscription made really simple to scale and carry the workloads from one site to another.
We never faced any major outage with the OpenShift platform. It's been running fine except for our network timeouts. That's it for this. It's not about, the pods themselves are running fine, but identifying this kind of issue hasn't been easy, but it didn't have any real business impact. It is just about other sre just to better serve our client and improve the platform's reliability.
