Wednesday, September 23, 2009

The 3-year CIO vs. IBM i: PHP to the rescue!

So are you staying on the IBM i? With downturned economies, abandoned maintenance renewals, layoffs and slashed budgets I wonder if you will stick with IBM i. We have all known for years that the TCA (Total Cost of Acquisition) for IBM i is not terribly competitive while the TCO (Total Cost of Ownership) is more in line with what you might pay for a real server or server farm. But the advent of server virtualization and more and more ubiquitous software in the Linux and Windows realm have made a serious dent in that philosophy. I wonder, are you staying on IBM i?

For years I have worked with the local user group in Chicago called The Omni User. Just in my tenure we have had countless speakers come from all parts of the IBM i community and geography. Folks like Alison Butterill, Al Grega, Jon Paris and Susan Gantner, Randall Munson, Larry Bolhuis and many, many more. Even I have picked up a session or two at a dinner meeting or one-day conference. But some years ago I recall a keynote at one of our annual conferences given by Dr. Frank Soltis. It was truly one of the more memorable presentations I have seen in a while. I must say I have gleaned quite a bit from his books and talks that I now know the folks there as well as my own kids! And, I rarely miss the opportunity to see him talk. One of my favorites was a COMMON opening session when he came out, placed his laptop down on the table, opened it and showed everyon in the audience the Apple logo on the front. The crowd went wild and his only response was: “of course you would expect I am going to run something reliable…”

At the Omni event, Dr. Frank was discussing the R&D group of IBM Rochester. He was talking about all the wonderful things being developed behind the scenes. All of it sounded exciting and then he made a point about Windows and Microsoft. The point he made was that IBM Rochester could not sit back and the could not sleep, that they absolutely HAD to focus on bringing new solutions to the market because Microsoft would eventually catch up. He said that Microsoft would eventually put out an OS that did not require daily reboots and weekly patches. Well, maybe he was half right. The folks at Microsoft have come out with better server software and now own the lion’s share of the server market. Dr Frank is a tremendous observer.

So I wonder, will you stay on IBM i? Has the Intel space captured you, even a little? Usually it starts out with a file server, something that is rather small and harmless. Then you discover the benefits of Active Directory and think “OK, What’s a couple more servers”. Especially when you consider how important authentication is to the organization you are told by EVERY Microsoft BP to “cluster” your AD. Then a specialty server or two, maybe even a SQL Server to satisfy the back end of a Sharepoint application. And then it begins. A user asks a developer if they can have the same data in SQL Server as they have on i5. The reasons are numerous and unnecessary for the purpose of this writing, but we have heard them all. Maybe you used DTS and moved to the new SSIS for replicating i5 data. You mused when the folks talked about how much more useful the data is here than on the i5. Then it comes, a new CIO.

I have joked around about the 3 year CIO in many presentations and writings. I have seen these individuals and they are numerous and plentiful. They are focus on a simple agenda: Year 1: Honeymoon and quick hits. Maybe they shake up the department with staff changes and a simple solution for a couple of key users in the organization. Usually loud key users are sought out at this point. During this time she is building up momentum and groupthink for the big project. Year two is the initiation of the big project. Something exciting like a new ERP would do the trick. Does the organization need a new ERP? That should depend on many things from a practical perspective. But we are not dealing with practical, here. From a functional aspect, a new ERP should be warranted when the needs of the company outweigh the capabilities of the software and a reasonable attempt by the IT department to keep up with the changes. Like a discreet manufacturer moving to process manufacturing, or something like that. But what I see more and more is the comment: “The AS/400 and the green screen are just not strategic…” In many cases I wish there was an IT police department I could call to have a restraining order put out on some of these guys. But, alas we cannot legislate stupidity.

Trevor, if you are reading, don’t send me any notes about the naming. I am just writing what I hear and see. And the 3-year CIO will never call it by its correct name. It is to their advantage to use the old name as they are trying to connote the “ancient” nature of the IBM i. It is a tumultuous ride during this second year and if all goes well, the 3 year CIO starts to prepare her resume for the next 3-year gig somewhere else. Certainly, there are variations on this where the CIO may last 4-5 years. But eventually they get bored and move onto look at a new challenge.

Can you hang on to your IBM i through the tenure of the 3-year CIO? I would guess that you wouldn’t. Primarily because the 3-year CIO is cost justifying all the new systems based upon the upgrade costs and maintenance fees of the IBM i. (TCA vs. TCO) And how many CIO's are reporting to the CFO? And what CFO wouldn’t be attracted to the smell of fresh cut costs? It’s like catnip to the tabby crawling around my feet as I write this.

Then there is the long term CIO who “gets it”. They understand the value of server consolidation. They appreciate the maintenance costs for the machine that really doesn’t need a full time administrator, in most shops. One of our customers has truly enjoyed thinking strategically. This customer went from a 3-year CIO to a more strategic CIO. The new CIO who took over from the 3-year CIO was presented with a contract to implement a very large German based ERP with a three letter acronym. This CIO wisely said “I’m not signing that without doing my own due diligence!” I love this guy, already!

The new CIO wanders down to the IT department where the news of the new ERP was causing the RPG developers to dust off the only useful Microsoft application at a time like this: Word - in preparation for the latest iteration of their respective resumes. He asks the IT guys “what’s up with this ERP project and what alternatives are there?” The CIO reiterated that the primary complaint about the existing system was the perception that it was old due to the prevalence of green screens. The IT guys looked at each other, dropped their resumes, and said “We’ve been playing around with PHP on a Linux box over there and I just heard that IBM is now supporting PHP on the i. The CIO gave the order to build and develop a pilot project.

The IT guys dug in and installed Zend Core on their System i at V5R4. In a couple of weeks they had GUI interfaces with data and charts and all kinds of demos for the CIO. Nothing earth shattering, but it was significant. The CIO was impressed and said “OK, How long to convert the whole shooting match?” The IT guys responded with ‘do you mean EVERYTHING?” The CIO nodded in the affirmative. So the IT guys headed back to the drawing board and developed an aggressive plan. A plan, by the way, that would take half the time and a fraction of the cost of the new ERP system. The IT guys not only built the plan, but they delivered the goods on time and just a little over budget (within 10 points). Try that with your ERP implementation!

The more I work with IBM i customers, the more I hear stories like these. PHP on IBM i is giving companies new hope and fresh approach to opportunities. Many of these opportunities have always been there with tactical tools and Java, CGI, etc. But PHP brings world class power within the reach of the RPG developer along with a roadmap that the CIO can fine tune to the needs of the organization. That is not to say that a new ERP might still be in order for some companies, but it does beg the question: Is your CIO on the three year track?
The moral of the story is that the new CIO should be embraced and not be feared. At least until year 2!


  1. This is so true. This is also true of companies that have no CIO and get direction from the CFO. So often the new guy wants to show off his prowess by changing something that does not need changing and we have to pay for it. He gets to put on his resume how he single handed put in this or that and how he cut cost. But at what price.


  2. Mike,

    This reminds me of something I was discussing with one of my friends from the OCEAN user's group (way out West in the land of sunshine and sushi). She and I were talking about what specific actions could be taken to fulfill the lofty objectives set forth by iManfest. The idea we came up with was to put together a task force whose only purpose is to contact IBM i customers who are thinking of leaving the platform. A sort of Seal Team 1 for the IBM i. These would be our best and brightest who would be on the front lines going up against the 3-year CIO.

    It seemed like a great idea because how many times have we been in that company with the 3-year CIO and had no allies? Personally, I've been there at least twice as an employee and probably two or three more as a consultant. It's a tough road.

    Alas, we come to the real most cases the decision to move away from the IBM i is not about technology or even money. It's about politics. The CIO has to make noise and look like he's doing something so that he can get his bonus at the end of year 1 and 2. It's tough to combat that.

    Great observations.

    - Jeff

  3. I am one of the 3 year CIOs and don't agree with the statements above. For one why would I want to run PHP on a $1 M server when I can get the same or better performance from a 15K server? Why do I need 3-5 RPG and COBOL programmers to do the job of couple developers? The big problem is that IBM has not gotten it. Hardware is now a commodity not a luxury, applications are readly available for every industry and we daon't have to re-invent the wheel with RPG and COBOL.

    2 years ago, I had 5 -570 IBM i systems; today I have only one and the only thing I am missing from my environment is the huge cost. My user community is happier, applications perform better and are real time vs batch. Can't wait to get rid of the last IBM i .

  4. To Mr-2-year-CIO-who-is-hiding-being-anonymity,

    1)Sounds like you were paying too much for your IBM i's to begin with, which might equate to you not doing your research on how you can get much more CPU for your dollar these days.

    2) Do you really think you can count it a win at this point with only two years under your belt? You have most likely just entered the cusp of code maintenance and haven't probably lived through many (if any) major language/framework/db/os updates/upgrades - all of which a non-events for IBM i.

    3) It is more than likely that your 3-5 RPG programmers were supporting A LOT of code, hence why they might have been lagging on developing new apps. You will experience the *same* thing (but worse) once your new environment reaches the point of maintenance. First round development ALWAYS goes much faster than making changes to existing code. Note also that you may have had dead-beat RPG developers (this is sadly not uncommon in the RPG space).

    Maybe you can post your name and your company so we know you aren't just ranting? Really, in the end, I want to know more details for your switch so I can know if you made a good decision or a bad one. Often times it is easy to get caught up in hype (I have done it) and miss out on what the IBM i was giving you. I don't disagree that IBM has sucked it up giving RPG programmers an RPG way to easily create nice looking GUIs.

    Aaron Bartell

  5. >in most cases the decision to move away from the IBM i is not about technology or even money.

    I don't know if I could agree with "most" cases, maybe "some". In short, IBM has NOT given their primary customer base (i.e. RPG shops) a reliable way to move forward that was guaranteed to not be pulled out from under then in 5 to 8 years time.

    I was on a webcast with Pavlok today and must say that I am seeing things in the Zend/PHP camp that actually stand the chance to be a good solution for RPG developers. Why? Well, probably the biggest one is that they aren't so political and over weight like IBM is. They (Zend) still have hunger to meet the community's needs along with make money off of the Enterprises - an en devour that requires a careful balance not to piss off either side yet provide plenty of paid/free functionality to keep everyone one board. IBM constantly tries to find that balance but usually ends up missing the boat because they come to the dock too late and have so many irons in so many fires that they have to compete with themselves.

    Notice the "Open EGL" stuff that is recently happening? Before "Open EGL" they thought they could introduce a new language to the world and have people purchase it, to the communities that are used to free frameworks, and thought it would take off like wildfire. Now they are trying to give it (an "express" version of the IDE) away in hopes of gaining popularity.

    Zend, my hat is off to you for seemingly being able to find a good balance.

    Aaron Bartell

  6. IBM has made such remarkable price/performance concessions in current Power Server line - to the point that I'm baffled that folks still raise the cost point.

    A new entry model 520 with IBM i offers 4,200 CPW at $6,381. That's 60 times more CPU power at about 32 percent the cost of an entry model 170 of 8 years ago.

    For comparison purposes, MS SQL Server Standard Edition retails for about $6K, while the Enterprise Edition retails for about $25K - per processor - and that doesn't even include hardware or operating system.

    My reason for stating it that way is because IBM i combines an enterprise-class database plus a world-class application server in just one box, at a remarkable price/performance point.

    Even if you look solely at cost to acquire, and completely set aside total cost of ownership, an IBM i system is a better value than a comparable Wintel solution.

    -Nathan M. Andelin

  7. Thank you Aaron for being willing to put your name on your post :-) I echo your sentiment to call out the 2 or 3 year CIO by name and company. If he has nothing to lose, why on earth would he hide? The IBM police are not going to come down from Rochester and steal your last i5! Glad you enjoyed the presentation. Lots more coming!

    I was tempted to delete his post but I will leave it there in hopes that other more honest folks would feel comfortable about sharing their experiences.



  8. Mr. 2yr-CIO,

    Having been a CIO myself, I am both surprised and disappointed by your post.

    If you paid 1 million dollars for an IBM i that you subsequently replaced with a $15k Linux/Wintel server then one of two things is true.

    Either you were vastly under-utilizing your IBM i


    You were robbed by your IBM sales rep.

    As to why you would use an IBM i over a Linux or Wintel server, the reasons are many but the most compelling are that the IBM i is much more reliable, scalable and supportable.

    Also, next time when you share a scathing opinion such as the one here man up and put your name on it.


  9. Could someone shed some light? based on the posts above I am thinking I have been robbed myself. Full disclosure: I am neither a CIO nor a Systems I fan and I am convinced based on my own experience that the system I is expensive, inflexible and outdated. We recently purchased an i570 with 12 TB of DASD; 9 processors and 128 GB or RAM and the cost of the hardware and licensing was ~ 1.2 M . At the same time we purchased an HP Matrix Blade System fully populated with 32 Procs (192 cores) 20 TB of SAN disk and 2 TB of RAM for < 500K licensing included. Was I robbed by my IBM sales rep? I like to think not as I consider him my friend. I also think that taking cheap shots to someone who perhaps has a lot to loose disclosing the company they work for is not a good argument to defend the system I

  10. Nardoni,

    IBM i runs on blade servers with SAN attached disk - if that's what you want to compare.

    On the other hand, if you really do want to compare the cost of IBM i Power Servers to HP Blade Servers, you should break out the software licensing and then run a benchmark suite to consider the price/performance.

    Regarding software, a 32 processor license of MS SQL Server Enterprise Edition would run about $800K, and Oracle licensing would be considerably above that - just for the DBMS. Then you can add OS and Hardware on top of that.

    Regarding the price/performance of a benchmark suite, it sounds like you're probably running different applications on the different servers, so any comparison there would be impractical.

    You probably can't make an apples-to-apples comparison. So why bother passing along a meaningless one?

    -Nathan M. Andelin

  11. Seems everyone here is feisty when someone criticizes the System I. I'll drop it but not before clarifying this point. I just licensed 312 MS SQL servers and cost for all was not even close to the 800K you state. I am comparing apples to apples. I can migrate the entire workload of couple system I-570s to a Blade matrix and have plenty of juice left over for many more systems and applications.

  12. Nardoni,

    I am not sure if your "cheap shots" comment was directed at my post or not, but the reality is you need to dicern cheap shots vs. holding accountable. If your company won't let you disclose who you are online then either post it as your own self (without your company name attached) or don't post at all. These comments are on the internet "forever" and the IBM i doesn't need any more bad press than it already has from people who we can't confirm are real.

    Of course, now we have your numbers to evaluate. I just got done purchasing an entry level IBM i 520 for $20K (if I remove my partnerworld discount). Here is the full config:

    Anybody have comments for that machine? I have RAID, DB2, IBM i, 4GB memory, compilers, ZendCore, etc on the config. My $20k number comes form subtracting all of the software I don't really need but added just because Partnerworld people get it for free and I wanted to check it out.

    Aaron Bartell

  13. Aaron

    Don't take me wrong, I evolved from the mainframe and was a big IBM fan. I recognize the great contributions and reliability of the AS400 midrange. Unfortunately I don't particularly care for how much the hardcore IBM'ers hate and degrade everything not IBM, I feel those individuals have a million reasons to stick with the 1980's technology and keep on paying for the 1980's prices. Yes I know, the AS400 evolved and the current System I has a lot going for it, but its not as cost effective as everyone tries to make it appear. I have been managing budgets and designing technology solutions for Fortune companies for several years and I speak from personal experience having acquired several IBM I's, having migrated away from it several apps and systems and TCO and ROI is just not something I have ever found on the system I.

    My comment wasn't directed at anyone in particular. As far as your 20K configuration, could you tell me how much it would cost to add interactive licensing and few TB of DASD? Can you tell me how much it would cost you to upgrade from 1 single core proc to say 4 quad or six core processors... you can't get that kind of machine for less than mid to high 6 digits in the system I. I just added 1 TB of DASD to a System I and it cost more than 20K.

    Software is another issue, I recently quoted commercial software and I had an option to run it on Wintel or the System I. For Wintel it costs ~ 75K, same software for the System I > 400K. Yearly maintenance for the Wintel software 14K for the Syetm I 90K. And if you dare go from a P5 to a P6 system I; you have to pay an upgrade charge on the software of 55K and maintenance increases to 115K. It doesn't matter if the workload is not changing, you are upgrading and that costs $$$$.

    As I told the sales person, upgrading to the newest hardware means a penalization on the software. Equivalent to you buying the latest hybrid car to get better millage and better reliability; yet when you go to the gas station to fuel up, you are told you have to pay $6.00 per gallon when everyone else pays $2,50; just because you are driving a more fuel efficient car. That doesn't happen in the distributed systems world. To me it simply a matter of principle.

    Lets face it, and as IBM puts it... the system I is the venerable machine of yesteryear. Today you can get a lot more for a lot less, that is the only point I am trying to make.

  14. Nardoni,

    Thanks for your balanced response. I don't disagree that the IBM i costs more. The thing we get to determine is if the additional cost gains us enough to justify it. It sounds like you can't justify the cost of an IBM i anymore and in that case you should migrate away from it. If you aren't running RPG applications then the IBM i starts to hold less and less value because all the other languages (i.e. Java/EGL/PHP/Ruby/etc) don't really make use of the IBM i integrated features. To be perfectly honest, if I wasn't heavy into RPG programming and instead exclusively did Java and PHP, then I would not use the IBM i.

    Nathan and I do a lot of RPG applications and subsequently can utilize the native features of IBM i to our advantage (integreated job management, integrated DB, integrated logging, serverside debugging, library lists, etc), all which make the IBM i a superior environment for developing and running applications in the long term *if* you can find a suitable new looking interface to program to (which Nathan has with his RelationalWeb framework).

  15. Nardoni,

    The other thing I wanted to point out in the last posting also is the fact that the gap that IBM once had and filled, compared to what other software stacks offered, is becoming much smaller. I agree that it is much harder to easily see the benefits the IBM i offers over others. For example, Microsoft/VMWare has found that by offer easy to use virtualization technologies that they can more easily cover up the fact that Windows isn't great for multiple big apps to all be running at the same time on a single instance. Instead you can now have a new virtual instance of Windows, preconfigured with whatever software you want loaded, setup in a matter of minutes (sometimes seconds), and the failover is getting better all the time.

    Just thought I would weigh in with those comments so you know I am not a bigot :-) Note though that if I didn't run IBM i for my server that I would instead run Linux and NOT Windows :-)

    Aaron Bartell

  16. Nardoni,

    You seem to be recalling "old" pricing terms for IBM i servers - such as a separate price for interactive features, which no longer apply to the new Power Servers.

    You didn't seem to quite catch my comment about IBM i running on low-cost blade servers and attaching to SAN-based disk subsystems - if that's an interest.

    Your pricing notions are out-dated.

    Are you serious about running 312 MS SQL Server instances? That would be an administrative nightmare and an army of technicians.

    -Nathan M. Andelin

  17. I'be been playing with VMWare today, but fail to see how it might be a cure for Windows woes.

    I understand the tendency to distribute workloads across multiple Windows servers to compensate for performance and stability problems. I also understand that VMWare facilitates that by enabling multi-core servers to be partitioned. But ...

    Folks end-up buying high-capacity servers, where more than half of the capacity exists for fail-over support. Is that less costly than buying a server that doesn't fail?

    Why do I say that more than half of the capacity exists for fail-over support? Well, at least one core is dedicated to the hypervisor. Then remaining cores are divided between other partitions - where generally half are set up for fail-over support.

    Then, in the case of web applications, you need to come up with some sort of load balancer, to dispatch requests across multiple partitions.

    Of course, each partition needs its own back-up and virus scanning procedures ...

    It seems to me that VMWare is overly hyped.

    -Nathan M. Andelin

  18. Just to clarify my comments about fail-over support, I was referring to software fail-over - not hardware.

    Under a partitioned server, a shared disk or network device failure would generally disable all virtual partitions - not just one.

    Virtual partitions complicate matters further in HA settings, as virtual partitions would need to be duplicated and managed at your disaster recovery site.

    It seems so much easier to just deploy applications under a runtime environment that doesn't fail.

    Folks are better off buying a 1-core IBM i server with 2 Gig of RAM, than buying & partitioning a 4-core Wintel server with 8 Gig of RAM.

    -Nathan M. Andelin

  19. This comment has been removed by a blog administrator.


Note: Only a member of this blog may post a comment.