In the Wheelhouse: A Cloud-First Deployment Model Is Unfair to Hardware Customers

  • Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times

In many of our environments, the "speed of cloud" is no faster than the "speed of APYPTF" or the "speed of RSTLICPGM."


That's right. I said it.


When I see a software demo or presentation that blows the doors off the current version, then I want it. Badly. Isn't that the point? Great presentation. Great demo. There's an engaging speaker. There's a buzz in the room and on Twitter. You can tell that your users would get immediate benefits when you load that new version. They're going to love this!


I caught the opening general session and a few other featured sessions at IBM Connect last week. I was so stoked with the changes coming to IBM Notes, IBM Connections, and a new redesigned take on email with a product called IBM Mail Next. IBM has done such a great job in designing a web-based email interface that on the surface Mail Next seems really great. I say "on the surface" because there was no deep dive available. The slides were not too plentiful, but what we did see looked darn slick. From a Gantt chart calendar view to a drastically simplified task list, Mail Next appears to be a good leap into truly making email look more social.


Then the strategy is displayed: cloud first.


Truth be told, it's not a major issue as it's probably going to be only a few months of a wait, but it is a valid point of contention for me. It irks me because customers who want to run the software on-premises have to wait. More importantly, the customers with a major investment in IBM hardware have to wait.


To be fair, I can understand a cloud-first model for a number of reasons.


First, I figure IBM wants to deploy to the cloud first because it's the fastest way to get customers (read: cloud customers) using the software. This drives conversation and adoption, and it allows them to tout SmartCloud success. IBM deserves to drive revenue like every other company.


Second, it reinforces public perception that cloud gets precedence. If I want the new stuff right away, then I need to be in the cloud, right? Does that drive new customers to the cloud? Perhaps some. But hey, reinforcing that perception certainly doesn't hurt. If I were looking to move to the cloud, I'd certainly give IBM's offering some consideration. However, I'm not looking to the cloud for a number of reasons based on my business needs and business rules.


The third point, and this actually makes sense from a pro-on-premises point of view, is that the cloud customers have given IBM a head start in ironing out the kinks in the cloud-based offerings before on-premises customers get a downloadable offering. I would suggest, however, that while any production cloud solution from a reputable vendor would certainly not be bug-riddled, it's quality assurance to some degree.


With that being said and the other half of a fair and balanced argument set, I (like many of you reading this) have chosen IBM as our primary hardware vendor. Given that the MC in MC Press Online stands for Midrange Computing, you're more than likely running IBM i, AIX, or PowerLinux on IBM Power Systems or maybe even PureSystems iron. You've made an investment in IBM hardware and software. If you're looking to get the latest and greatest updates and support from IBM, then you're most certainly paying software and hardware maintenance costs.


In addition to standard IBM i and Power Systems maintenance, I'm paying for support on a number of IBM software products, and I'd expect as a paying customer to be held in the same regard as those who've gone the cloud route. The fact that this software is running on IBM hardware drives that point home a little further. IBM i on Power Systems customers especially, with the borderline fanaticism of our community and devotion to the operating system with rock solid hardware, are a grand bunch who fly the IBM flag, defending it in budgeting deliberations, dark alleyways, and the like.


I've also been a long-time advocate of consolidation and maximizing utilization out of existing resources. Many Power Systems customers have workload to spare in terms of idle processor cores and idle memory, allowing customers to add capacity when the business requires. Think about it. I usually use my shop as a small-to-medium business example. In 2002, my company had an IBM iSeries model 270 with 1070 batch CPW, 50 CPW interactive, and one processor core. This model 270 was one of those Greenstreak promotion machines that IBM sold a ton of. Today, we have a Power 720 Express with 5900 CPW out of the box and one core activated when we bought it just a few years ago. The Power 720 is the most popular Power7 system in the wild, so we're not out of the ordinary by any means. Most Power7 customers have a Power 720 with one or two active cores. Depending on the model, there's a 4-, 6-, or 8-core processor, so chances are there's plenty of extra horsepower. For many shops, the processing capabilities have far exceeded their processing utilization over the last 10 years. For my shop, that's an absolute certainty. In the span of 18 months, we added enough workload to justify turning on one of the idle five processor cores, effectively doubling the processing power of that machine. The beauty of it was that we tracked the CPU usage over that time with IBM i performance monitoring to see a steady climb. All we had to do was assign the core to the partition and then IPL in order to activate it. Of course, we paid IBM for it too, but the second core was almost like getting a brand new server in terms of performance.


This is a major selling point of cloud computing that's built in to our existing hardware investment! The fact that IBM i requires such little attention from information technology staff allows us to focus on adding business value. We've got such an amazing platform that when I see advertisements touting the value of cloud computing, I usually think, "Well, we already do that. It's just not called cloud." I guess you can call it private cloud, but for me that's even a stretch. It's just IBM i and Power Systems.


I'm thinking we hardware customers deserve equal treatment. Whether it gets deployed right away or not is the customer's decision as we all have our own schedules, resources, and business rules. Some of us have the luxury of being able to power down for a few hours once a week on a Saturday night in order to do PTFs if we so desire. Some of us are more liberal in our fix pack updates because we see the value in keeping bleeding-edge current and can tolerate minor downtime or have a clustering solution in place to ensure that most upgrades will have zero downtime.


Too often I hear from many cloud advocates that the average IT department doesn't have the resources or skills to compete against the cloud. Those arguments are usually broad-brush fluff pieces entitled "5 Reasons Why Your IT Department Can't Keep Up" or something ridiculous like that. Did global IT have a perceived sudden drop in skill level right around the time cloud computing started to fly in the press?


In many of our environments, the "speed of cloud" is no faster than the "speed of APYPTF" or the "speed of RSTLICPGM." long as we're able to download the software first.