What is our primary use case?
I work for a manufacturing company. They use it for the B2B landscape. They use IBM Sterling Integrator for both the EDI transactions and managed file transfers. We have Sterling File Gateway as well as the Integrator for translation.
We have two different environments to handle, and we have close to 12,000 suppliers and carriers. They do EDI transactions through us using Sterling Integrator. We facilitate the EDI transition to the respective SAP equivalent data translations, whether it’s an IDOC or a flat-file, based on the requirements. We also onboard the Sterling partners on the system, whoever wants to do EDI.
How has it helped my organization?
When it comes to automating file transfer in a business process here, it has been automated-- a user drops the file into the mailbox, and it just auto-triggers as per the schedules to execute the series of steps that we would have written in that workflow. That is how automation is conducted.
We need not worry about the file delivery because it takes care of itself. If there are any errors, we have an error handling mechanism defined within the workflow itself, which details out on which step the error actually occurred, what the error was, and things like that.
There are some common errors which we can find out through the Business Flow logs. It is quite self-defined on its own with respect to Sterling.
What is most valuable?
I really like its offer of a system and business process that handles inbound and outbound transactions. That feature I like which is built. We can do customization on top of it.
Beyond that, the way the mailboxes are defined and the way we can connect to them and trace down the EDI file is a feature that I like.
In comparison to any other tools, the UI is very user-friendly to understand and the way it has been categorized.
What needs improvement?
I’ve been part of this client for the last seven-plus years. It’s been close to 70 to 80 continuous improvements we have delivered. The priority ones which we always shortlist are the recurring incidents or recurring issues, which came in the initial phase of the year when we took this entire landscape under our maintenance.
One such incident I can recollect is with respect to performance tuning. We committed to our users 99.99% and above as the availability metrics for Sterling Integrator. This has acted as a high-availability system, but we treat it as mission-critical. When it comes to the commitment we give to users, we have to ensure the system is kept most stable.
So, the majority of the problem was in the communication channels. Whenever we enabled additional logging for the communication channel, the system used to have hiccups.
So we worked with the vendor, stating that the visibility channel framework needs to be changed because the moment we enable more logging, it literally brings the system down, or the system doesn’t work as it should. They took our input and delivered a better framework in their next releases, which helped us after upgrading to have that stability intact. As the system grows, we ensure to have performance tuning triggered and optimize the business process wherever required.
For example, by default, Sterling Integrator business process will have full logging enabled. We took care of those things. Not all business processes or workflows require full logging enabled. Only a few critical ones require every step logs.
For the rest, we categorized and reduced the logging for those workflows. That actually helped us to increase the IO overall from ten milliseconds to six milliseconds. That was a good achievement.
Apart from that, in terms of queues, how we maintain the queues, how we defined all file queues across the critical business process is one thing we felt was done better. The threads we assign for the priority queues and the business processes were configured to those priority queues, whatever is critical, so that it gets high priority to allow the threads to process. So that queue thread Sterling was taken under the performance tuning.
Apart from that, I think some of the best practices which IBM recommends is what we usually run through every year. We just have the health check done through IBM, and we just ensure that all the best practice recommendations are added in the system.
For how long have I used the solution?
For the last seven-plus years, I have been working on Sterling Integrator. My overall experience in the integrations field is close to fifteen years now.
Overall experience in the integrations field is close to fifteen years now.
What do I think about the scalability of the solution?
Scalability depends on the underlying business needs. So if the system requires adding a number of cores, we can scale it as per the requirement.
For example, as of today, we support close to three to four million transactions per month. For that, we have a set of underlying infrastructure built around, which might be 12 cores in each node, good enough to throw out some 70 threads for these transactions to be handled. And we still have some room to occupy for more volume increase. The moment we feel that there is more volume coming in on a monthly basis, it turns five million.
That’s when we have to think about scalability. As of now, there is vertical scaling happening in the current environment, but we can go for horizontal scaling.
I would rate scalability an eight out of ten. Whenever there is a need to add additional resources and scale it, we touch every section. For example, we touch the database, the underlying Linux, and any file system. We also ensure network feasibility is available. So, I would rate it eight out of ten.
There are no major challenges, but there are a few hiccups in terms of connectivity. Whenever we do performance tuning, things usually go right. The only challenges we face are with CPU utilization or memory allocations. We might have configured too little or too much based on the current system availability. When the CPUs or file monitors are overloaded, it crashes. So, those are the factors that pose problems during scalability.
How are customer service and support?
We have ongoing vendor management with IBM for any system-critical issues or questions about features. We ask if we can enable certain features and if we need to take care of anything before enabling them. We also contact them for system-down scenarios. We have an IBM portal where we open a PMR, and they assist us. Most of our questions are about upgrades, vulnerabilities, and system-down scenarios. As part of the n-1 version maintenance, we follow a trend of upgrading to the latest version to use all the features IBM offers.
Whenever we do any upgrade, we tend to encounter some issues. These can be known issues or new ones we discover for IBM.
We have a few hiccups here and there in terms of the people who support IBM. It probably depends on the person who is actually taking care of the PMR.
We have very good support members who are very knowledgeable, but we also have a downside where a few people are probably new to the team and do not understand the criticality.
But ultimately, whenever the ticket goes to the engineering team, we always have the answer.
How would you rate customer service and support?
Which solution did I use previously and why did I switch?
I’ve been part of multiple integrators, right from SAP Process Integration to SAP PO Process Orchestration. I began my career on SAP and then moved to IBM product, which is Sterling Integrator. So, I have knowledge of multiple integration tools.
What's my experience with pricing, setup cost, and licensing?
The product is not cheap. With respect to pricing, IBM is in the top two or top three after SAP.
It is worth it. But it definitely depends on the environment and the requirements. So it can change. IBM can be suited for multiple scenarios or environments, but it may not be suitable for a few as well.
What other advice do I have?
I would definitely recommend it to other people. I would definitely recommend it in terms of the user interface that IBM has today and the support model they have because I have not come across such a good support model in different integration spaces.
Overall, I would rate it an eight out of ten.