Cloud Foundry Reviews

The solution is used as a runtime for microservices developed in Node.JS for a SaaS mobile solution.

We had +/- 20 microservices which serve an app developed for Android and iPhone. The microservice has a direct connection to MySql, PostgreSQL, and identity services. 

The solution is hosted in the IBM Cloud Dallas region, and the clients (mobile app) are in Peru and South America. Most of the users use just 4G to access the services. The use case doesn't require high availability of the services, and it was enough not to invest in redundancy.

The administration of the runtime environment is almost zero, as most of it is being managed by Cloud Foundry itself. Very easy to auto-scale applications both up and down and good observability features. It almost eliminates the admin role. 

Cloud Foundry and IBM Cloud also manage the network and security settings, so network and security senior specialists are not required to set up the infrastructure. 

Deployment of new instances is also very fast, so a very complex CI/CD environment is not required to apply changes to the code.

We enjoy the fast deployment. Cloud Foundry builds the runtime environment directly without requiring dependency management from the developer or administrator.

The autoscaling is great. It is just a switch that needs to be turned on, and autoscaling starts working. At this moment, you begin to see different meters about usage that helps you in updating the scaling limits, which help you tune the running instances. Besides this, autoscaling can be scheduled, so in times of low activity, you can have lower limits or increase in advance for special dates. 

It has good logging. CF has logging events that help identify when a transaction runs and its response time which helps in monitoring execution. 

In IBM Cloud, the product has been deprecated in favor of Kubernetes, which is a more complicated infrastructure to manage.

CF should be merged with the Kubernetes project. This could benefit both projects in areas of easy deployment and manageability. Cloud Foundry could include multizone support so it can be run in an enterprise production environment.

Another consideration is the typical "the chicken or the egg" problem. Should you run containers on top of Cloud Foundry, or should you have a Cloud Foundry container? If this is more clearly defined, a long life for Cloud Foundry can be secured.

I've been using the solution for three years.

The stability is good.

The scalability is good.

Customer service was good until the announcement of deprecation was released. then problem resolution took longer. 

Positive

Before Cloud Foundry, I ran workloads in virtual machines.

The initial setup was very easy as I used the IBM Cloud service.

We have developed a SaaS solution and currently we are concern more in acquiring additional customers with a low entry-point prize and having an personalized support. The trick with SaaS solutions is to increase the volumen of customers and democratized/demonetize your solution. 

Use autoscaling to define the right number of instances. Usually, the cloud providers define a default size for memory or the number of instances. Try to see if you can decrease these numbers so that, perhaps at the beginning, you don't need those overestimated sizes.

I did not evaluate other options as I started with IBM Cloud and there the entry point was Cloud Foundry. I heard from Heruko long time ago but at that time I was not thinking in a cloud environment.

Look for a cloud provider that has long-term support for Cloud Foundry. IBM is deprecating it.

The aim was to use the smartphone for an IoT project.

Our objectives were to show how sensor data generated by the smartphone could be sent, stored, and visualized using the MQTT protocol.

Environment consisted of following prerequisites:

• Ubuntu
• Node-RED
• Node.js
• Web service 

We made use of the IBM Watson IoT Platform, a cloud-hosted service, instead of building a platform from scratch. Then, we created web apps on IBM Cloud Foundry that processes, visualizes, and stores the data.

We also had to design the flow in Node-RED.

It improved in following ways:

1. Improved installation time: Given the number of prerequisites, the setup would have taken a long time, especially because of the fact that I had a Windows PC and was using Hyper-V. I would first have to setup a VM environment. 
2. Integration: Integrated modules are available seamlessly, like mobile, database, and authentication.
3. Interface: Node-RED graphical flow diagram.
4. Ease of use: Node.js for hosting IoT components.
5. Minimal support: Using Bluemix (IBM cloud Platform-as-a-Service (PaaS)), it provides a platform by which multiple application components (e.g., database, authentication, etc.) can be implemented quickly.
   

The ability to focus on the application and service by working around and by passing the technology stack.

Node.js is the most important advantage as it provided us with the ability to start rapidly using a pre-built solution instead of having to deploy from scratch and install a lot of prerequisites.

My favorite component of IBM's solution is Node-RED, which greatly shortens the amount of time required to develop, test, and deploy new applications.

I feel that IBM is still primarily for techies as far as the user interface is concerned. Since there are now guys coming from non-IT backgrounds, I feel that some improvement is needed compared to AWS, but more specially compared to Azure.

After the initial excitement period with Node-RED is over, you crave the need of additional integrations to third-party services.

Two years.

Stable during the time of implementation.

You are allocated a minimum amount of resources in the free tier. This seems fair and highly scalable, as you pay per usage as per cloud pricing schemes.

No.

By moving to IBM Cloud Foundry, we reduced our project time by a factor of two and decreased costs by a factor of four because it was a PoC.

IBM has a free tier and payment option depending on the products selected.

The pricing models should be reworked to the needs of a wider range of companies. Some customers will not be able to afford it until quite a few years into production, even after good PoC results and a successful launch.

I evaluated Azure and AWS, comparing the three of them. IBM is the only vendor to offer integration with blockchain for smart contract development (compared to Azure and AWS).