What is our primary use case?
We chose OpCon to replace a scheduling package that was controlling approximately 10,000 batch jobs every day. So the main purpose of OpCon, for us, is to replace an aging homegrown solution with a more advanced scheduling product that has more bells and whistles. We use it for job control. We have Enterprise Manager on desktops communicating to agents that are on our mainframe computer.
We haven't yet completed the conversion. We are about 30 percent converted right now. We still running 70 percent of the work through our old scheduling package. We have two main shops. One of them is an upstate shop and one is a downstate shop. I run the downstate shop. We have about 10,000 jobs, of which 5,000 to 6,000 are in that downstate system. We have deployed about 2,000 jobs out of a total of 6,000 jobs, downstate.
How has it helped my organization?
The part that jumps out is the notification process. The agent can now notify us, by email or text messages, when any jobs have failed or when any groups of jobs have finished successfully. Previously, it was a manual process where somebody would say, "We finished the work now," or, "A job has failed," and then they would have to start sending out emails or calling people to notify them when we received certain errors or reached certain stages in the work. That part has been automated.
We anticipate, in the future, that it will save us time mainly because, with the old scheduling package, we would have to manually identify and calculate dates for the next 12-month period. We would have to do that every single year. That's a very lengthy and accident-prone area and, by automating, we expect to see a reduction in effort from the staff.
What is most valuable?
There are three features which are valuable:
- automated calendar functions
- the notification process for failed jobs or unscheduled events occurring, via email and text messaging
- the ability for the scheduling package to communicate across multiple platforms.
We have three mainframe computers and our previous scheduling package wouldn't communicate across the mainframes. OpCon gives us that ability to schedule jobs on mainframe A and a job on mainframe B and the latter can be dependent upon a job on A.
Those are the key components that we've found to be beneficial.
What needs improvement?
There's a large learning curve which, for some of our less technical staff, has been an issue. It's still new to us. Every week we're finding new ways of doing things with the product. What we miss the most is having an in-house expert whom we can call upon every single day. Literally, every single day, I or my staff have to go to the documentation and work out how a certain function works or why it reacted in a certain way. And that can take a lot of time and effort. But what has been beneficial is having SMA's 800 number which we call if we can't work it out ourselves. But many times we try to work it out ourselves rather than calling them up five to ten times a day.
We're converting 200 jobs at a time or 500 jobs at a time. We'll find out, once they're in place: "Oh, wow. There's a better way that we could have done that." And then we have to go back a little bit and figure out if we should have done it this way or scheduled it that way. It's a very powerful tool and we're not always choosing the right choice the first time through, when scheduling our work. That's why we miss having somebody onsite to say: "No, you really shouldn't have done it this way." We're actually finding out sometimes the hard way.
The calendar interface and the frequency interface is a very powerful, yet complex, section of OpCon in which all our staff have made mistakes. They have implemented what they believed was logically correct and then afterward discovered that their logic was flawed because OpCon did it a different way. That part, which is incredibly useful, is also incredibly dangerous. The interface or the ability to directly do more functions within the frequency definitely has room for expansion. As good as it is, it can be a lot better.
For how long have I used the solution?
It was first installed in 2018 and we started using it for production work at the beginning of 2019, so we've been going for 10 or 11 months.
What do I think about the stability of the solution?
The stability has been very good.
The downside is that when something does go wrong, most times it's a networking issue, which tends to get lost in the mix. OpCon will say, "Unable to communicate," and now we have to try and track which part of it has failed. Is it the agent that has failed? Is it the Enterprise Manager that has failed? Is it the network backbone that has failed? Or is it the SQL Server that has failed? A way in which OpCon could be improved is to better analyze things when a failure is occurring to point us in a better direction without our having to check all the different paths.
What do I think about the scalability of the solution?
I love the idea that we can scale it, but what I don't like is that every time I consider wanting to scale it to something else, it costs a lot of money and then I have to jump through hoops with all of my hierarchy in order to get it. So it's good and it's bad. I actually haven't seen any scalability yet because nobody has approved the enormous amounts of money that are needed to put another agent in another area.
We have about 24 active users and their main function with OpCon is purely to monitor and schedule the work on the different platforms. What I would like to see happen in the future, and I know this does exist, is to expand the user group to the client base or to the development group so that they can then see the results of their work in a read-only manner. Because we're concentrating our efforts on deployment, I haven't yet gotten around to getting that part implemented.
Ideally, I'd like to see three people on it on every shift to monitor this amount of work. Their role would be to monitor the workflow, to implement new applications into OpCon, and to ensure the frequencies and calendars are working as expected. As good as OpCon is, we still need to verify that it's interpretation of when we've told it to run the jobs actually matches up with what we really expect it to do. We just don't trust it completely yet.
How are customer service and technical support?
Technical support has been excellent. We had two people from SMA who were part of the project to do the conversion. Now that they're no longer available to us we miss them tremendously. But we also understand that they had to move on to other projects.
What has been beneficial, and I have no complaints about, is that every time we do encounter a hurdle of any kind, when we call the 800 number, whatever technician we speak to at the other end is extremely knowledgeable and walks us through it. But the hard part many times is that they don't necessarily know how we are set up so there's always that 10 or 15 minutes as we explain, in our terms, how we're doing business so that they can understand what it is that we could have done better or what we're doing wrong. Having an in-house expert would be extremely beneficial but that's too costly.
Having a dedicated tech from OpCon, about three months ago, would have been extremely beneficial. We used up an awful lot of the time and resources of the dedicated people who were assigned to this project when we weren't even fully aware of the questions that we were going to ask because we hadn't implemented anything yet. We had them available to us during a stage when we were still putting all of the jobs into the test system and not into the live system. That's just the way it worked out. And again, when you're trying to convert so many jobs that are mission-critical, it's very difficult to take the risk of it not working correctly, so we're being very cautious about how we implement all of our work.
How was the initial setup?
The initial setup was complex. We had a training course that was given to us back in August but almost everybody who attended the course didn't actually get to use the product, hands-on, for about six months after the course. Nobody could really fully comprehend OpCon when we were first given the course. It was a very different product to what we were used to using. As a consequence, it was like a brand-new language and many of our staff couldn't wrap their heads around. It's not until you actually use it that you start to understand how this thing works.
Our deployment is still going on. I would say it's been a 12-month deployment with about another three months to go before we complete it. We're anticipating having it fully deployed by February of 2020.
The first part of the implementation was that we took a flatfile database dump of our current scheduling product and that was provided to SMA support, to Kevin Adams and Ben Adams. They loaded that into the OpCon database. Then we would project future schedules within OpCon and compare them to future schedules in our in-house scheduling package to see if the conversion had gone as expected. Once we found all of the different nuances, the different parts that had been interpreted incorrectly — meaning either their schedule dependencies or frequencies, probably because we exposed to them wrong — the next phase was to do parallel running.
We continued to run all of our work in our existing scheduling package and each day we would run the same schedules in OpCon but convert all of the jobs in OpCon to null jobs so that they performed no functions. They wouldn't start anything. They would just run and hopefully run in the same sequence as our live system.
The third phase was to actually start the conversion. We identified the least mission-critical jobs, the low-hanging fruit which were the least damaging jobs, and converted those. We turned them off in our in-house scheduling package and turned them on within OpCon. Once that proved to be successful, we then broke down jobs into groups to be converted, initially starting out with groups of about 100 to 200 at a time.
We've now reached the final phase, which is the remaining 3,000 or so jobs. It's a very complex schedule. We were going to implement it in stages and we're finding that it's very difficult to implement jobs that are running it OpCon while still running our old scheduling package when we have dependencies between them. So the final phase is proving to be a little bit more daunting but we're getting there.
After deployment of OpCon, it took about two-and-a-half to three months to automate our first process, between when it was communicating with the agents on the mainframe and when we actually started to run jobs.
What was our ROI?
It's too early to tell about ROI.
Which other solutions did I evaluate?
We evaluated Control-M from BMC. Both OpCon and Control-M were going to provide us with the solution that we were looking for. The decisions were then out of my hands because it was then left up to the money people. The final selling point was that there was another state organization that was already using SMA. I believe the Civil Service Department is using SMA. That was the final factor: If we were going to purchase something, let's try and keep them looking the same.
What other advice do I have?
I would highly recommend an onsite evaluation of OpCon that has already been deployed and seeing it fully in action, so that you could be better prepared to ask the right questions prior to getting it. All we saw was a remote demo and that, to me, was a big mistake on my people's part and probably SMA's part. We never got to see it in action so we didn't know all the right questions to ask.
My biggest lesson in using OpCon is that I wish I'd been more involved at the beginning of the project, when they were estimating the need for support. We should have budgeted for a different type of support during the early days.
The second big mistake was that there is a latest and greatest version of OpCon, which I believe is called OpCon Deploy, and we didn't budget for it or know of its existence until after we were doing our deployment. That would have made such a huge difference, because everything that we were doing in our deployment was manual: We had to extract the information from our scheduling package provide it to SMA support. They would manipulate the data, put it into our test system, and then, to roll it across from our test system to our live system, they would have to export the database or export the schedules and import them into production OpCon. Whereas Deploy is fully automated. That would have made a huge difference. We didn't pay for it because we weren't told about it and as a consequence, this is what we got.
We still wish we could get it but now we can't get it because we have to wait for the budget people to approve it. And to get the budget people to approve it, we have to give them the same explanations as when we were going from our old scheduling package to the new scheduling package and they're not buying it. They're saying, "No, no, you already used that as a reason for us spending a half a million dollars. You can't use it again."
Right now, I'm going to rate it as an eight out of 10, but I believe it's going to be a 10 for us.
Which deployment model are you using for this solution?
Which version of this solution are you currently using?
Accelerate digital transformation through workload automation
Automate repetitive tasks so you can focus on projects that drive your business forward. Find out how OpCon workload automation enables you to create repeatable, reliable workflows - all managed from a single platform.