Apache Kafka alternatives and competitors

We're using Anypoint MQ for messaging, in different types: from SAP to third party systems, global third party systems, etc. There's a hundred of them.

It's extremely expensive to change things in Anypoint MQ, e.g. we have to go through either Accenture or IBM. IBM will do the changes, but the actual change we ask for costs huge amounts of money. I'm not sure why that is, or whether that's just a contract, but they always complain about multiple changes, regression testing, etc. It sounds like it's a very long winded task just to make a simple change. This is an area for improvement for Anypoint MQ.

Another problem we have with Anypoint MQ is that some of our end systems cannot cope with the throughput. If we put too many messages down to it, then there's a delay to send those messages onto the system that can't take them in as fast, and this means that the queue blows. We have either a 40 gigabyte or 40 megabyte queue space, but if we send 7,000 messages which need to be combined and sent out, the queue blows, because it just can't manage that throughput because the output is so slow. With the memory, we don't have a big enough queue to cope, so there's that problem.

If there was a way around that problem, we could send it. What we need to do is to be able to send 50,000 messages an hour or more, which we can do SAP to SAP, but we haven't found a solution on how to cope with sending it to Anypoint MQ because it blows up, so that's the big problem we have.

What we'd like to see in the next release of Anypoint MQ is for the issue of slow output of messages to be solved. Even if we wrote into disc, e.g. the disc space is cheap, so if you could write the messages to disc and then just feed them out and when the queue is caught up with itself, that would solve our problems. It means we could send MuleSoft all the messages in one go, and then it could drip feed them out.

The additional feature we'd like to see in Anypoint MQ is to be able to write the message queue to disc, to allow for ample queueing of messages, because currently, the output to a particular system is slow.

We've been dealing with Anypoint MQ for 20 years now.

Anypoint MQ has been stable, and it has not caused us any problems.

Anypoint MQ is a scalable solution.

We have not contacted the technical support team of Anypoint MQ, because we have our own internal third parties to deal with any issue that could arise from the solution.

What we've noticed in Anypoint MQ is that it's expensive to change things.

We evaluated Apache Kafka.

I thought we had a problem with our queuing software, and it turned out that it's not the software itself, it's the third party. We're passing the messages on to the third party, and they can only receive it at a certain rate, so we had to drop the Anypoint MQ software down, but it's not actually the MQ software, rather it's the actual third party application that was causing the problem.

As I'm working on the SAP side, I'm not sure if I'm the best person to discuss about the features I like most about the solution.

My advice to others looking into implementing the solution is to first look at what you're trying to achieve, because you've got Anypoint MQ as your queueing software, and then you've got the Apache Kafka software as well, and they do different things. I didn't realize this when I started looking, because I was looking at Kafka thinking that would be the answer to our problem, but it wasn't, because it looks great if you want instantaneous messaging and loads of bandwidth, but you need to do the interpretation of messages.

I'd say, look at what you want it for, because Anypoint MQ seems to be as good as the other options, to be fair. I'd also tell them about the problem I've had with it, and that's not just going to be a problem with Anypoint MQ alone, because I think other systems will have the same problem, e.g. the queuing problem. I'll tell others to really, really look at the application. Anypoint MQ seems to be a standard application, and it's acceptable.

My overall impression of Anypoint MQ in terms of rating it, is seven out of ten, but that's just based on my feeling. I don't have any quantifiable reasons on why I'm giving it that score.

We are all using the file transfer or MQ FTP feature. We are also it for distributed queuing and clustering.

Currently, we are not using many advanced features. We are only using point-to-point MQ. I have previously used features like context-based authentication, SSL authentication, and high availability. These are good and pretty cool features. They make your business reliable.

For critical business needs, everyone uses only IBM MQ. It is the first choice because of its reliability. There is a one-send-and-one-delivery feature. It also has a no-message-loss feature, and because of that, only IBM MQ is used in banking or financial sectors.

It would be an advantage if they can include streaming in IBM MQ, similar to Kafka. Kafka is used mainly for streaming purposes. This feature is clearly lacking in IBM MQ. If they add this feature to IBM MQ, it will have an edge over other products.

I have been working with IBM MQ for the last 14 years.

IBM MQ is a very stable product. You also get very good support from IBM, but we rarely have to go back to IBM for support.

It has good scalability. We are using point-to-point or distributed MQ, so we are not that much worried about scalability. If we need scalability, we can use MQ clustering for a high workload. We can configure it for resiliency and high availability by using the multi-instance queue managers. If one of the nodes goes down, it will automatically failover to the other node. It also provides some advanced high availability features on top of the multi-instance queue manager.

You get very good support from IBM. If you are facing any issues that are tricky or there is any code issue where FDC files are being generated and you're not sure what is happening, you can open a case with them. They will help you with that. They are very efficient.

The initial setup is very simple. The installation doesn't take more than 15 or 20 minutes.

I have installed it myself. I'm also doing maintenance, patching, upgrades, and migrations. We have a team of 11 administrators who are working on IBM MQ. They use it on a daily basis.

The upgrade process is simple. I refer to IBM Information Center. As a part of the preparation, I go through all the steps that they have given. I correlate the information with the infrastructure that we have. According to the current infrastructure, we document the requirements, and after that, we do the upgrade. We couldn't do in-place migration or upgrade, so we had to do parallelization. We took a new server, installed the new version, created a new queue manager, and migrated all the services.

It is a licensed product. As compared to an open-source solution, such as RabbitMQ, it is obviously costly. If you're using IBM Message Broker, which is a licensed product, IBM MQ is included in the same license. You don't have to pay separately for IBM MQ. The license cost of IBM MQ is lesser than IBM Message Broker.

I have been asked to do a PoC for one of our use cases, and we used RabbitMQ for that. They wanted to assess RabbitMQ in comparison to IBM MQ.

Obviously, IBM MQ has more advantages when compared with RabbitMQ. The main reason for doing this PoC was that RabbitMQ is an open-source product. Cost-wise, it looks effective, but from a technical point of view as well as from the point of view of scalability and features, IBM MQ is very enriched.

I would definitely recommend this solution, but it also depends on your needs and business case. I have been using IBM MQ for the last 14 years. I am very much used to it, and I like it. I have used other products too, such as RabbitMQ and Kafka, but not that much. 

I would rate IBM MQ an eight out of ten.

We are in the early stages of evaluating Red Hat AMQ for an OpenShift container platform because it can provide a very good Kubernetes platform using asynchronous data communication.

The solution is very lightweight, easy to configure, simple to manage, and robust since it launched.

There is improvement needed to keep the support libraries updated.

I have been using this solution for approximately one and a half years.

We have found the solution to be reliable.

The solution is highly scalable. This is the main reason why we are using this solution.

The support technical documentation is very good.

The initial setup is very easy.

There is a subscription needed for this solution and there are support plans available.

I have evaluated Kafka. Red Head AMQ solves the producer-consumer problems and Kafka is used for streaming mainly.

This solution is very mature and can be very useful depending on the use case. For what we use it for, it has worked perfectly. For some other use cases, this solution might not be the best to use. Most of my clients are using other Red Hat solutions combined with this one, such as Ansible and 3scale.

I rate Red Hat AMQ a nine out of ten.