Settling In With IBM i For The Long Haul
February 11, 2019 Timothy Prickett Morgan
If nothing else, the IBM i platform has exhibited extraordinary longevity. One might even say legendary longevity, if you want to take its history all the way back to the System/3 minicomputer from 1969. This is the real starting point in the AS/400 family tree and this is when Big Blue, for very sound legal and technical and marketing reasons, decided to fork its products to address the unique needs of large enterprises (with the System/360 mainframe and its follow-ons) and small and medium businesses (starting with the System/3 and moving on through the System/34, System/32, System/38, and System/36 in the 1970s and early 1980s and passing through the AS/400, AS/400e, iSeries, System i, and then IBM i on Power Systems platforms.
It has been a long run indeed, and many customers who have invested in the platform started way back then and there with the early versions of RPG and moved their applications forward and changed them as their businesses evolved and the depth and breadth of corporate computing changed, moving on up through RPG II, RPG III, RPG IV, ILE RPG, and now RPG free form. Being on this platform for even three decades makes you a relative newcomer.
There is a longer run ahead, since we believe that the companies that are still running IBM i systems are the true diehards, the ones who have no intention of leaving the platform and that, at least according to the survey data we have been privy too, are intending to continue investing in, or even expand their investments in, the IBM i platform.
Thus far, we are not in a recession and heaven willing there will not be one, so the priorities that IBM i shops have are not the ones that they had a decade ago during the height of the Great Recession. Back then, as was the case in just about all IT organizations, IBM i shops were hunkering down and were trying to cut costs in all ways possible, including deferring system upgrades and migrations as well as cutting back on other projects. Only 29 percent of the 750 IBM i shops that participated in the 2019 IBM i Marketplace Survey, which HelpSystems did back in October 2018, were concerned about reducing IT spending. This is a remarkably low level, and I think is indicative of how relatively strong the economy is – excepting some of the fits and starts we saw at the end of 2018 and here in early 2019 that make us nervous and could start putting pressure on things. Here are the top concerns as culled from the survey:
Dealing with the growth in data and in figuring out the analytics to chew on that data ranked a little bit higher on the 2019 IBM i Marketplace Survey than did reducing costs, and I think over the long haul these issues will become more important than modernizing applications and dealing with the IBM i skills shortages that are a perennial worry. Both of these issues are being solved as new programmers and new tools to make new interfaces to database applications are becoming more common and as technologies such as free form RPG, which looks more like Java, Python, and PHP, are being more widely deployed and, importantly, can be picked up more quickly by programmers experienced with these other languages.
Given the nature of the customer base, it seems unlikely to me that security and high availability will not continue to be primary concerns, despite the fact that the IBM i platform is among the most secure platforms on the planet (and not just because it is obscure, but because it is exceedingly difficult to hack) and it has a range of high availability and disaster recovery tools (from IBM, Syncsort, Maxava, and HelpSystems) available for those who want to double up their systems and protect their applications and data. The bar is often higher than simple backup and recovery for many IBM i shops in the banking, insurance, manufacturing, and distribution industries that dominate the platform. These companies can’t have security breaches, and they can’t have downtime.
There is a remarkable amount of stability in the IBM i customer base that we think, at this point, is reflective in the stability of the IBM i platform and Big Blue’s own belief that it needs a healthy IBM i platform to have an overall healthy Power Systems business. We all know that the Power Systems hardware business has just turned in five quarters of revenue growth – something we discussed recently in developing our own revenue model for the Power Systems business – but what we did not know, and what you should know, is that in the second and third quarters of 2018, the IBM i portion of the business grew significantly faster than the overall Power Systems business, and the only reason that this did not happen in the final quarter of 2018 is that sales of IBM i machinery in Q4 2017 was quite strong and represented a very tough compare. The point is, the IBM i business has been raising the Power Systems class average. (These hints about the IBM i business come compliments of Steve Sibley, vice president and offering manager of Cognitive Systems at IBM.)
IBM’s own financial stability of the Power platform – which has been bolstered by a move into Linux clusters for analytics and high performance computing simulation and modeling as well as by the adoption of the HANA in-memory database by SAP customers on big iron machines including Power8 and now Power9 systems – helps IBM i customers feel more confident in investing in the current IBM i platform. The recent evidence from several different surveys, not just the one done by HelpSystems every year, suggests that companies are by and large either continuing to invest in the platform or even in some cases are planning to increase their spending on the IBM i platform in 2019.
As you can see, the pattern of investment plans for the IBM i platform, as shown in the chart above, has not changed very much at all in the past four years. It is a remarkably stable pattern with but a little wiggling here and there that may not even be statistically significant. Just under a quarter of IBM i shops have reported in the past four years that they plan to increase their investment in the platform in each year, and just under half say that they are holding steady. This does not mean that the same companies, year after year, are investing more and other companies are staying pat, year after year. It is far more likely that every handful of years – more like four or five – customers upgrade their systems and expand their capacity, and they then sit tight. The wonder is that the split isn’t showing far fewer companies investing and far more sitting tight. That more than a tenth of the shops don’t know what their plan is as each prior year comes to a close is a bit disturbing, but it is honest and shows that a significant portion of shops have other priorities aside from hardware and operating system upgrades. We have said this before and we will say it again: We think that the people who respond to surveys and read weekly publications focused on the IBM i platform are the most active shops – the ones more likely to stay relatively current on hardware and software. So the pace of adoption for new technologies, and the rate of investment, should be higher than in the actual base, much of which does not change much at all.
So if we had to adjust this data to take on the whole base, there might be far fewer sites that are investing more money, far more companies that are sitting tight, and maybe fewer sites that are contemplating moving off the IBM i platform. I think the distribution of data is probably something like 10 percent of shops have no idea what they are doing investment wise with IBM this year, 5 percent are thinking about moving some or all of their applications to another platform, maybe 10 percent are investing more this year, and the remaining 75 percent are sitting tight. This is just a guess, of course. As far as we can tell, the rate of attrition – how many sites we actually lose each year – just a tad over 1 percent. So the rate of movement of applications off the platform, or incidences of unplugging IBM i databases and applications, may not be anywhere near as high in the overall base as the data above suggests. What is alarming, perhaps, is that the rate of moving some or all applications off the platform is balanced against those who say they will increase investments. Perhaps these are hopeful survey takers, and those who think it is easy to move find it is not and those who think they will find the money to invest will not.
What we do know is that if the rate of application attrition was anywhere near as high as these surveys suggest, then the IBM i business would not be growing, but shrinking. And we know it is not shrinking, so we think there is a disconnect between planning and reality, both on the upside and the downside.
If you drill down into the data for the 2019 IBM i Marketplace Survey, there were 13 percent of shops that said they would be moving some applications to a new platform, and another 9 percent that said they were going to move all of their applications off IBM i. (This number is consistent with the recent ALL400s survey done by John Rockwell.)
Anyway, good luck with that.
Porting applications from one platform to another, of buying a new suite on that new platform, is an exceedingly difficult task. It is not like trying to change a tire while driving down the road, as is a common metaphor, but rather like trying to take the tire off one car moving down the highway and installing it on another car driving beside it in the adjacent lane without crashing either car or smashing into anyone else on the road. Optimism abounds, but when push comes to shove, very few companies try such a maneuver, and when they do, it is usually because there is a corporate mandate, more times than not caused by a merger or acquisition, that pits some other platform against IBM i running on Power Systems. Companies that say they are making such a move off IBM i are sanguine for their own personal reasons, perhaps, but they are not necessarily realistic about how long it might take, what disruption it will cost, and what ultimate benefit, if any, will be realized.
If you do the math on the chart above, eight-tenths of the base has no idea how long a move will take, another 1.7 percent thinks it will take more than five years, and 3 percent say it will take between two years and five years. Only 3.4 percent of the total base say they can do it in under two years. We think all of these numbers are optimistic, and the companies who could easily leave OS/400 and IBM i already did a long time ago and those that are remain have a harder time, not an easier time, moving. If this were not true, the IBM i base would be a hell of a lot smaller than the 120,000 customers we think are out there, based on what Big Blue has told us in the past. This is the difference between fear or pressure or culture and the reality of trying to move a business off one platform and onto another. These moves are always a lot harder than they seem on the front end, and we suspect many of the benefits also don’t materialize for those who do jump platforms.
At the average attrition rate suggested by this survey data – 9 percent move off the platform in somewhere between one year and more than five years, with most companies not being able to see more than five years into the future that is a neat trick – the installed base would shrink dramatically. It is tough to say how far because of the wide range of timeframes in the survey. If it was 9 percent of the base within two years – call it 4.5 percent of the base per year – then within a decade the overall base would shrink from 120,000 IBM i sites worldwide down to about 72,000. This would dramatic indeed. But at a 1 percent attrition rate per year, the base is still at 107,500 unique customers (not sites and not installed machines, both of which are larger) by 2029. We think there is every chance that the attrition rate will actually slow and drop underneath 1 percent as IBM demonstrates commitment to the Power Systems platform and its IBM i operating system. There are always some new customers being added in new markets, to be sure, but the bleed rate (even if it is small) is still probably an order of magnitude larger than the feed rate.
When they do think about making the move, IBM i shops know exactly where they want to go, and this answer has been gradually changing over the years: Linux as an alternative to IBM i is on the rise and Windows Server as an alternative is on the wane. In the latest survey, 52 percent of the companies that said they were moving all or some of their applications to another platform said they were choosing Windows Server, while 34 percent chose Linux. This reflects the relative popularity of Windows Server and Linux in the datacenters of the world at large, and may be tipped just a little more heavily towards Linux compared to the rest of the world. Interestingly, 10 percent of those polled who said they were moving were looking at AIX platforms, and another 4 percent were going upscale to System z mainframes – as unlikely as this may seem. Platforms tend to roll downhill; they do not usually defy gravity like that.
The thing about such surveys is that they show intent, not action. We often intend to do a lot more than we actually can accomplish, and moving platforms after spending decades of building up expertise is not usually a very smart move unless the platform is in real trouble – like the Itanium systems from Hewlett Packard Enterprise running OpenVMS or HP-UX or the HP 3000s running MPE or the Sparc systems from Oracle running Solaris. These were once great platforms with huge installed bases and tremendous revenue streams, but now, IBM is the last of these Unix and proprietary platforms with its Power Systems line. And it is by far the biggest and for sure the only one showing any growth.