The Many Layers Of IBM i Managed Services
October 26, 2020 Timothy Prickett Morgan
(Sponsored Content) One size rarely fits all across datacenter sizes and industry types, and the needs of the IT department change because conditions in the business change over time and across geographies, too. And that is why it is important to find a managed service provider that has both wide and deep experience with any particular platform.
Datanational’s Managed Services division is one of the early players in the IBM i space, and while its origins are in the American Midwest, its current operations spread out across the United States and span the globe from there. We recently profiled Datanational’s high availability and disaster recovery business, and this week we have an in-depth conversation with Sandi Djokovic, vice president at the company where he has worked for more than a quarter century. Datanational was founded in 1979 and began as an IBM midrange reseller that moved into managed services, which became a large component of Datanational’s business.
Timothy Prickett Morgan: People might not be aware of this, but managed services have not been available from a lot of suppliers in the AS/400 and IBM i market for a long time. It’s a relatively new market by IBM midrange standards because so many customers took care of their own systems and applications for decades using System/3X and AS/400 gear.
Sandi Djokovic: Things are different now, that is for sure. The whole IT landscape has been changing, evolving over the years. There are so many more players out there compared to a dozen years ago, when if you were to write an article about IBM AS/400 hosting providers, I think you’d have a super short list. We would have been on that list back then, of course. We have been doing this officially with IBM since 2008, which is going back to IBM’s MSP program roots.
What brought us into the MSP fold here was we had a super large account that we had been working with for years who basically said: Everybody who manages and maintains the system – including the hardware and operating system – we are basically not going to have them on staff. They asked us to come in and take over. So our Power MSP practice developed from the humble beginnings of one client.
TPM: Why did this customer not just go to IBM? Big Blue has been doing outsourcing for a long, long time.
Sandi Djokovic: Well, actually, they did have IBM for some hosting before us, but under the IBM arrangement, the client was still doing a ton of hands-on stuff. So they took a phased approach where they used to have their system on site, then IBM spun up a system for them but was still not doing any kind of 24/7 monitoring and managed services.
When they approached us, they wanted a full-blown solution provider. And even though this was a fairly large client, IBM Global Services was not built to scale down to that level and they did not do full management where they could actually monitor and support the system 24×7. IBM was ringing a bell when something happens and the client still needed a person on the other end to do something about whatever the issue was. And this client just wanted to jettison those responsibilities and focus on application support.
We had to shift into research mode and figure out where we were going to put the system. We had our own data center, but it was certainly not high enough of a grade where we were comfortable guaranteeing five nines uptime at that time. We still had to sell this client a full system, too, even though we were only managing it for them. The whole MSP program from IBM was in its infancy and IBM had to work through how we, as a Business Partner, would play the role of a customer all of a sudden. That’s one thing that has changed over the past 12 years: our revenue model. We used to be just traditional reseller and now we hardly resell anything to clients. We still have a few, but we are the customer for IBM most of the time, and this relationship has had to evolve.
Our challenge back then was staffing up. We always had a lot of hard-working people, but we certainly didn’t have people staying up 24×7 watching systems and responding to messages unless it was an emergency we were called in to. We had to figure all this out.
TPM: Well, you clearly did. One of the things I want to clear up is the difference between buying managed services and buying cloud capacity. This is an important distinction people need to understand. Managed services is more than a personal cloud in that it augments or replaces some of the IT administration staff.
Sandi Djokovic: The emphasis with us as an MSP is providing a private secure data center environment. There are some IBM i cloud solutions that are real now, instead of being hypothetical or just a testbed for developers. Today, you can spin up instances on IBM Cloud, Google Cloud, Microsoft Azure, and Skytap – but that’s not our business.
We’re from the Midwest, and our client base is for the most part in manufacturing. We do serve clients globally, but many of those accounts are North America manufacturing companies that have overseas operations. But the underlying theme there is they want to know where data resides. Everyone likes the cloud concept when it comes personal data like iPhone contacts and photos and such, but when it comes to enterprise data, customers want to know exactly where their data is located and how it is secured.
We deploy infrastructure on our IBM servers, with our IBM i operating system licenses and our IBM licensed program products and do it with our people in North America, who also manage our firewall and we provision Internet bandwidth and monitor the servers. It’s all within our facilities, and it is known where it is and it does not move around.
TPM: Choosing a managed service provider partner is not like renting some generic X86 server capacity and some storage on a public cloud. Is this priced like an outsourcing contract, with a three-year term and monthly fees? This is the kind of commitment somebody makes for a long time.
Sandi Djokovic: If it is a decent size system, as you might imagine, it is difficult to write a short-term commitment because it requires an investment on our part. We have done some short-term contracts, for instance when clients are just moving into legacy mode. But for the most part, we do three-year to four-year initial contract terms and we typically renew annually after that.
TPM: Let’s say I have a typical IBM i installation. I’ve got a couple of sockets worth of IBM i running. I’ve got a reasonable amount of storage. I have probably hundreds of users to thousands of users, depending on how heavy their workloads are. It’s at a company that makes a couple ten million dollars a year, sometimes a couple hundred million dollars. When they decide to make that first phone call to you and then they decide to pull the trigger, what are the steps they have to go through to do this transition to buying your managed services? And how long does it take to do this kind of transition, assuming they have not really moved anything to the public cloud as yet.
Sandi Djokovic: It depends on a lot of these factors that you mentioned. Our sweet spot is not really playing to multi-billion-dollar corporations. We do have some of those big companies on our roster as a client, but it’s typically divisions of those companies. You may have a large company that, you know, over the years have acquired different companies around a bunch of different ERPs, and they have this $750 million annual sales division that runs on the IBM i platform with 500 users.
We have a checklist of things that are part of the onboarding process that we go through – it’s really a full-blown, red carpet service that starts with an assessment of its current environment. We look at the sizing of the system, and this is a little tricky because a lot of times, customers have some big behemoth of a machine but are using only a tiny fraction of it. You don’t want to oversize the solution, and obviously price yourself out of the deal or charge them too much for what they need. At the same time, you need to understand the elasticity that they might need in the future.
There are also application software license key and application ownership considerations we have to take into account. If you migrate to a machine that’s no longer in the enterprise and is at an MSP model, does that change anything? With some application providers, that makes a difference. For the customers who have invested significantly over the years into their ERP systems, our goal is that they maintain the ownership of any perpetual licensing they have paid for.
This is one thing that has changed now in the past dozen years. Back then, it was a lot more difficult to work with some third-party application providers to have them accept the transfer of a customer license product to an MSP machine. There was a lot of old guard out there, and they just wanted to sell new licenses to clients, to get a $50,000 repurchase of a product versus maybe a $2,000 transfer fee. That’s largely gone, and it is very similar to how ISVs do things with IBM i when it comes to disaster recovery workloads.
TPM: How long is that MSP migration process? Does it take a year? Does it take a year and a half? Does it take two years? Four months? You have to do the assessments, really understand the environment, then buy a system or find capacity for it in your datacenter, then move the applications over and then get the users over in such a way that it’s transparent. That seems to be the tricky part.
Sandi Djokovic: On these more involved systems, such as our initial customer, it would be a four-month to six-month timeline. That said, we have done system migration’s in two weeks, but that was back it up to our datacenter, restore it there, sign in, make sure that the client could validate data, and then do the real cutover the next weekend. We like those when we get them, but they don’t happen most of the time.
The larger deals require planning because we’re going from something older to something newer, so we’re changing more than just the serial number. Moreover, we go through multiple iterations of migration because the first one is done in test mode, and then we cutover. The wild card factor there might be the OS change. We work with the client, understand their level of risk tolerance and funding. As you know, the more that you want to minimize risk, the more it’s going to cost and the longer the timeline because of the increased effort by all parties involved.
TPM: Do you get any demand from customers who say: I can’t buy a Power9? In fact, I can’t even buy a Power8, and I need a Power7 because that is the last hardware level that supports the operating system level where my ERP system is stuck.
Sandi Djokovic: We have had some of that, but interestingly enough, it doesn’t happen as much anymore. We can find V5R4 systems and provision for customers if we need to, but the issue there is still software licensing. In most cases, with any kind of legitimate ERP package, they’ll need to have a license key that can be used on a system with a different serial number. And if they’re in that position, they likely discontinued maintenance support on the ERP system years ago and they don’t have access to new license keys. So there is a negotiation, perhaps to obtain a one-time arrangement for the transfer to happen. And in that case, we always try to implement them on the newest OS version as a can be supported on their application just to provide as long of a timeline as possible.
TPM: Do customers sometimes come to you because they’re moving to a different ERP system and they want to set up the new ERP in an MSP setup to begin with and then get help do their conversion from the old ERP system? Do you often get that double whammy, or is that just too much change for most companies to cope with?
Sandi Djokovic: That would not be like a common occurrence. We do see migrations where somebody is going to SAP and they’re not going to deploy it on the IBM i
TPM: Last question. I have a certain amount of capacity and I want to an ERP system and I am thinking of the public cloud or an MSP service like the one that Datanational has. When you buy an instance in the cloud, you got to manage the OS and the applications, and that takes people. My basic assumption is that it’s going to cost a little bit more because that work the MSP is doing does not come free. But on the other side of the IT budget, companies no longer have to pay for admins to do it. How do you compare the MSP offering in terms of ERP compute and storage equivalency compared to doing it on premise in general and doing it on reserved instances in the public cloud?
Sandi Djokovic: That’s a really good question, and comparing MSP to on-premises is really the common exercise for us to go through.
We actually go through the exercise and put together a matrix that says here’s what our MSP solution is going to be and how this compares to buying a new server and the cost of maintenance, electricity, data center space, security, and other costs. Periodic OS software upgrades and PTF maintenance are in there. And of course, service level agreements to the users needs to be calculated. And so on. It is a rich matrix.
We stack it all up, and if we are not at 15 percent to 25 percent savings while upgrading the level of service compared to on-premises gear, we tell the customer there is something missing in the data they are giving us. We save them money and deliver better service, and we keep them from being in the position of being on outdated, unsupported hardware on an old release of the IBM i operating system.
As far as the cloud goes, over the last three years, we have come maybe across not even a dozen instances of this kind of exercise. And every time the client comes back and says to take the public cloud option off the table. I think that the early cloud options for IBM i were just not scalable down to a level of, say, 30 to 50 users and they still had to support the environment themselves.