The idea behind cloud computing used to be simple – let the cloud service provider provide and manage the IT services and infrastructure while you get on with your business. However, reality got in the way and cloud computing has become yet another dimension in the complex web of IT service delivery. It is against this backdrop that the announcement of IBM Cloud Satellite at this year’s THINK conference needs to be judged.

Public Cloud meets the real world

The public cloud provides great flexibility for developers - allowing them to obtain the computing resources they need when needed and for just the length of time that they need them. When I was a VP of software development, much of my time was devoted to obtaining the computing resources my teams needed and for this the cloud has been a great boon.

However, it is now clear that many organizations are not willing to move certain types of data into cloud services because of compliance concerns. Not all workloads can easily be moved into cloud services for a variety of reasons including connectivity as well as configuration. Although the public cloud may be ideal for development it may be difficult to move workloads developed in one vendor’s cloud to another vendor’s one or back on-premises.

This has led to the growth of today’s hybrid IT delivery model where some IT services are delivered through the public cloud, some through private clouds, and others remain on-premises. Far from making things simple, the multiple IT service delivery models with multiple service providers have added to the cost and complexity of managing and securing business-critical systems upon which the organization depends.

No single approach meets all the challenges

One approach to these challenges is to use a common environment, such as VMware, for both on-premises and cloud delivery to provide a common management model. This is good if your organization has already chosen that environment.

However, that does not solve all the problems of connectivity. In some cases, the systems need to operate during periods where connectivity is limited or non-existent. These include maritime solutions during the time at sea as well as geographically remote operations such as mining. In other cases, low network latency that can only be obtained through physical proximity may be essential. These include manufacturing and logistics solutions.

In response to these challenges some cloud vendors now offer a form of “cloud in a box”. This is often an appliance that provides a service compatible with their cloud, but which can be installed wherever the customer chooses. This allows software developed on their public cloud and data or components with special requirements to be deployed in the most appropriate physical location. The service provided on the appliance may also be managed by the cloud vendor in the same way as the rest of the cloud service.

IBM Satellite Cloud

On May 6th, 2020 IBM announced a preview of IBM Cloud Satellite which they claim will deliver ”IBM Cloud services anywhere a client needs them, delivered as-a-service from a single pane of glass, through the IBM public cloud.”

This is based on Red Hat OpenShift which is a hybrid cloud, enterprise Kubernetes application platform. IBM Cloud Satellite extends the IBM cloud using the concept of Locations which are places outside of the IBM cloud where the customer can run applications. The customer can populate a Location with a set of Hosts (RHEL Linux machines) that run the cloud services. Locations are connected to the IBM Public Cloud through Satellite Link which provides administration and application-level firewall management and network security capabilities. In principle, this will make it possible for any service in the IBM Cloud Catalog to be run anywhere including in Satellite Locations.

It is important to note that this is an announcement and not a GA release. Many of the details are yet to be worked out, for example, the minimum infrastructure requirements. However, it is based upon the existing Red Hat OpenShift which is GA and is widely used.

IBM says that this offering will initially be based on infrastructure provided by the customer. At a later date, it will likely be extended to use IBM Private Cloud from IBM GTS. IBM is also working with hardware partners on an IBM Cloud Pak System Appliance.

This offering is clearly biased toward supporting Kubernetes-based development and container-based services. This is widely used as an approach to the development of new applications and application modernization. However, the recent announcements of Red Hat OpenShift virtualization extend the platform to allow VMs to be migrated to OpenShift without the need to fully containerize, enabling them to power mixed applications and be containerized over time (or not).

However, some customers may want to use other environments such as Cloud Foundry. IBM says that its goal with Satellite is to enable a broad set of IBM Cloud PaaS services to run on client-defined locations. However, the initial focus is on extending the IBM Red Hat OpenShift on IBM Cloud and Data services.

Is this a game changer?

From IBM’s point of view, this announcement makes sense given their acquisition of Red Hat. It also moves them towards the most popular approach to modern application development based on Kubernetes and containers. The recent announcement of OpenShift Virtualization widens the scope beyond this and, depending upon how widely it is adopted, may in the future be seen to be a game changer. However, IBM Satellite Cloud is still well behind the existing GA offerings from Microsoft (Azure Stack) and AWS (AWS Outposts).

See also