The world of enterprise IT is always in a state of perpetual change. The one thing that doesn’t change is the ongoing change we are all living with and occasionally challenged by. What’s maybe most interesting about the change going on today is that the change is not one major shift but many different shifts. Some of these are interrelated and overlapping. Others less so.
One of the major changes has, somewhat obviously, been the shift to cloud. This shift is ongoing and is very broad, encompassing both public and private clouds, and indeed multi-cloud. There are some other changes going on which are distinctly related to this cloud move, and these are very much an enabler to this change. This is the ‘containerization’ of IT deployments, seen today in the widespread adoption of Docker containers and Kubernetes environments to deploy them. This technology underpins much of the public and private cloud use, and itself is driven by, and enabled through the shift to a ‘devops’ style of management which allows for better use of resources, and for businesses to be far more agile. This approach is described in a number of ways. Typically this has been as “cattle and pets” to describe the difference between containers and more bespoke systems, but one of my colleagues (thanks Woz) likes to describe the environment as more like hire cars, versus your own car.
With a hire car, you start using it when you want. You do what you want with it. Then when you finish using it you stop using it, not thinking about it again. You haven’t needed to maintain it. You haven’t done anything to it. That car may as well no longer exist as far as you are concerned. Compare that to your own car. You decide exactly what car you want, then once you have it, you likely have it for a long time. You might personalize it. You probably maintain it, and enhance it over time with new tyres, wheels, exhaust. You leave your things inside it, knowing that when you go back to it, they will still be there.
That analogy is pretty good for the difference between containers and more traditional long-running environments. Containers are stateless. Once you get rid of them, there is nothing left in the container. This is like a hire car. When you pick up a hire car, it is exactly as you expect it. Nothing inside it. And when you return it to the hire car company, you better remember to take your luggage, your sunglasses, and the rest of your family members, as they certainly won’t be in the car if you come back and hire it again the next day or the next week. The car you own, however, has state. If you leave a bottle of water in it, then it will still be there the next time you use it. The fuel level will be the same (unless your children have used it and emptied the tank). And your sunglasses will be there within easy reach when the clouds clear and the sun comes out.
So, let’s talk about IBM MQ and containers. Because MQ is, at its heart, a stateful product. It preserves state in the form of messages. And yet if containers are stateless, why would you run MQ in a container? Isn’t it a contradiction? The answer is no. But it is certainly something you need to think about. And that goes back to why, and how you are using containers. As mentioned above, you are using containers probably as part of a devops environment. You will be deploying applications in containers, which will run as long as needed, and then the container, and the application will be removed. At least until next time. But what does IBM MQ do? It connects applications together. It provides a long running persistent environment to allow multiple applications to reliably and securely exchange messages. It doesn’t matter to MQ if one application in a container goes away. MQ just sits there and runs. It waits for the next application to appear and to put and get more messages. Some messages will sit in the queues for longer than others, depending on the message and depending on when the consuming applications are running. MQ, in essence, doesn’t really care if it is running in a container or not. MQ has supported containers since 2015. MQ can be run natively in Docker based container environments, in Kubernetes environments, in Red Hat OpenShift and in IBM Cloud Private. Indeed the recent MQ on Cloud hosted service is deployed as MQ in containers on both IBM Cloud and AWS. But in many expected use cases, although MQ will be running in a container, it is unlikely that the devops plan will see those MQ containers brought up and shut down as frequently as application containers are brought up and shut down. The administration team will need to be sure that all the messages in the queues have been drained before removing the container running MQ. Otherwise they will destroy a message, which is likely to have business value.
While you may run MQ in a container, businesses should be aware those containers are likely to be much longer running, because MQ is stateful, and preserving that state means keeping MQ up and running.
In summary, you absolutely can run IBM MQ in containers, and in your choice of container environment, such as IBM Cloud Private, or Red Hat OpenShift, or a combination. With a container based devops environment, that might be the best way to deploy and manage MQ. And there is a new way to license MQ running in containers as described here. However, the long running nature of MQ might also lead you to review whether, if MQ might be running continuously for months or even years, whether you want to treat MQ the same as your stateless applications. Does running in a container really make sense? It should certainly be thought about. You might even consider deploying MQ in a VM or maybe even deploying the MQ Appliance, which you could even think of as a container – just one that is rather more substantial than the ephemeral nature of the other containers you are using.
Many of the updates that IBM has made with IBM MQ over the last few years have been focused on responding to the customer choice. Wherever and however customers run applications, IBM MQ will be there to support those deployments and environments. On the cloud as a managed service. In containers. As a physical appliance. On the mainframe. Meeting your needs. Never losing a message. No wonder it’s hard to contain your excitement.
Next steps might involve downloading container images here.
Or reading more about MQ and containers here
