Is Upgrading a Silly Waste of Time and Money?
March 12, 2007 Hey, Brian
We currently have an AS/400 Model 720 at V5R1 and it runs perfectly fine. Two years ago, our business partner told us we were at the end of the line with the Model 720, which has with 512 MB of memory, is rated at 225/35 CPWs, and has 105 GB of disk (after RAID striping) on nine 4 GB drives and five 16 GB drives. We are not at 80 percent of disk capacity. I couldn’t find any good reasons to listen to him. I still can’t see any good reason to upgrade my software from V5R1 to V5R2, V5R3 or V5R4.
Moreover, we’re only growing our disk space at about 10 percent a year. That means that the model 720 has two or three more years left in it before I need to look at anything else. Why should I upgrade my software? Why should I upgrade my hardware? I think it’s silly and a waste of money.
Maybe you are right and maybe you are wrong. One thing for sure, the cost of being wrong is a lot more than the gain of being right. Let’s take a look at this in more detail.
Six or seven years ago, that AS/400 Model 720 would cost about $110,000 without software. The cost range of a System i that can do the same job for your company is between $40,000 and $120,000 and offers more disk, more memory, more CPW (from 600/60 to 3400/60). Compared with your current system, any of the i5 520 options are blazing fast. One of the tricks in the single-level storage architecture of the AS/400, iSeries, and i5 servers is that information is stored in the middle of every disk platter, and then as information is added, it is saved on either disk of this initial track. This minimizes the disk seek. Consequently, as more and more data is spread over the disk, the average seek time increases since the arm needs to travel farther on the average. Whether you notice it or not, your system is slowing down.
The slow disks on your current system add more drag on the system than the storage they provide. Today’s 15K RPM drives with the cache controllers not only store more data, but they are substantially faster in all aspects. Jeb, quite frankly, that Model 720 should be sent to the bone yard and it should never be rescued.
Among the reasons that you have been able to run your business and did not have to upgrade over the past two years, as you explained it to me, are these:
1. Other than routine program maintenance, no major applications systems were scheduled for development and no package systems required additional resources.
2. The defined Web applications were either outsourced or there was no Web development at all.
3. You did not undertake any advanced development applications that needed features available on later versions of the operating system.
4. V5R1 is not supported by IBM. V5R1 has been out of date since September 2005. V5R2 is out of date as of April 7, 2006. So are out of time here, too. If you have a software problem and you need help, IBM will tell you to upgrade to a supported release before you call back in, or they will charge you a Consult Line fee to get you back to functional. The end of support date for V5R3 has not been announced, but IBM will stop selling it on January 8, 2008, and if this is when V5R5 comes out, then V5R3’s end of life will also be announced. Additionally, all other things being equal why would you want to be two or three releases away from the current version when new IBM support personnel are trained only on the new versions.
5. The Model 720 hardware is still supported by IBM and will be supported until IBM finds it difficult to reclaim parts from machines returned from lease. You may find maintenance costs rising prior to this. When IBM discontinues hardware support, it usually provides customers with several years’ notice. Notwithstanding that you have a maintenance contract, the disk drives that are spinning are old. You have RAID, so this should not be as much of an issue. But with disk drives that are six or seven years old, when one disk goes, I guarantee that during the period of between when a disk crashes and when a new drive is being loaded, you will be sweating bullets.
6. Though the 720 system passed the 60 percent storage threshold (the point at which it is good to order more disk drives), performance has remained adequate for the last several years. Moreover, the system has not begun warning operations personnel that there is a critical shortage. The 720 is now above 80 percent utilization. When the system hits 90 percent, it will tell the operator that it is at a storage threshold. When there is not enough space to create and extend the temporary disk files to operate, the system will stop and will need a manual power down. This would be a very undesirable situation.
If there is not enough workspace to IPL, the system may not come up. It may need to be booted from tape or CD or through DST and special procedures to help free up operating space. In the worst case, the operating system and the back up may need to be restored from tape. After the arduous task of freeing space, the system should come back up and be ready to be used at the high storage level. However, the system will have reached a critical level and you will have to walk on eggs (use extreme caution) so as not to exceed the limit again.
Your hardware upgrade options are minimal to non-existent with IBM. Big Blue no longer upgrades the 720 system, but you can upgrade it yourself with used components from third parties. Make sure you find out what that does to your maintenance contract. As you know, this is not a good situation, and though at 80 percent capacity a system stop may not be imminent, it is a situation that may very well occur within the next year.
Why Upgrade–The Bottom Line
With the 720’s inability to run the current release of the operating system and the pending announcement of yet another release (V5R5, maybe this year and definitely next year), the storage situation getting close to the system threshold, and the relative slowness and old age of the current system, it is time to make the change to newer and thus more reliable hardware technology and to upgrade the OS to be able to more readily accomplish your new application requirements.
I recommend that you install a brand new System i5 520. I think you would be tickled for a number of years with an i5 520 Turbo model. Over the period in which the 720 has not been upgraded, IBM has evolved the system several generations from the AS/400 to the iSeries to the System i–as you well know from reading this newsletter.
The platform is now fully equipped for modern applications including e-business, relying less on its characteristics as a batch and transaction processing server. It also has the ability to run Windows, AIX, and Linux as well as Domino groupware.
With the introduction of the System i5 generations of machines, IBM has cut system prices, in some cases, to almost half off for specific models, making a great incentive to purchase up-to-date hardware, operating system, and utility software releases. IBM and its business partners encourage customers to stay current with updates to new releases so that the system is current and thus easier to manage. If a shop cannot be on the current release, IBM recommends being no more than one release away from the current release i5/OS V5R4 while addressing the factors limiting the ability to remain current.
Software upgrades for System i or any other platform basically do two things. They offer your business new and/or improved features, and they fix bugs (especially design bugs) that are present in older versions. Whether your company should upgrade most often depends on your need for the new bells and whistles or the performance improvements. It also depends on whether you have been experiencing regular problems because of software bugs. On some systems and periodically with System i, your budget comes into play if you must first acquire and pay for the upgrades. Most releases of System i software come bundled on the system– if you already have the license.
Software updates may not always be called updates or upgrades, but these are the popular terms today. Like it or not, there are few people in and out of IT who can avoid the curse of having to keep their software up to date. For example Nokia cautions its cell phone users: “Keep your software up-to-date or fix a potential operation problem you may have by updating your phone firmware regularly.” Firmware is low level software that enables the hardware to function, but it is software nonetheless. How much more important than your cell phone is it to keep a system such as the IBM System i, with millions of lines of code, up to date. Hey, so what if the phone has to reboot? What’s the big loss when all you have to do is open and close the phone flap? Fully capable computer systems, of course need a little more finesse and a lot more work.
There are a number of questions that need to be considered when you are exploring a software upgrade:
Will the upgrade offer you needed capabilities that you currently do not have? Because System i software has advanced so much in recent years, the list of new functionality is exhaustive. In fact, it is so exhaustive that many IT professionals have not been able to keep up with all of the new function because they must first do their jobs.
In each new release, there are more than several hundred new features that have the potential to change how you use your System i. Before making the decision to deploy, you need to discover how these innovations may work together (hopefully seamlessly) to give the company a more enjoyable and functional platform. Perhaps an upgrade would permit secure Web capabilities or provide a no-nonsense means of deploying your current applications on the Web. But do you really need that capability?
Will the upgrade enhance your ability to share files securely with co-workers, vendors, or the government? Over the last several years, the operating system has been enhanced with a more functional Integrated File System (IFS) and an enhanced peer file/print server called NetServer. This provides an integrated method of sharing files with people who use the Microsoft’s Windows and its System Message Block (SMB) protocol. Each year, its capabilities and its ability to secure directories, files, and documents has been enhanced. If you won’t be able to take advantage of the new facilities that previous versions don’t offer, this would not be a compelling reason to upgrade–at least not yet.
Are you being disrupted by software bugs or design deficiencies? If you are concerned by software bugs that affect the product set in the version you are currently using, you might consider upgrading if the new version adequately corrects the problems. Past implementations of newer functions such as the WebSphere Application Server have been problematic and often difficult to work with. Moreover, the short term fixes can create more problems than they solve. Newer versions of WebSphere, however, are much better than the old stuff. They are much more rich in function, easier to use, and easier to fix when they go awry. Though new versions may not fix all bugs and that they sometimes introduce new ones, the overall experience is that software improves over time and major chunks of fixes are provided in new OS versions. If there are no such bug issues in your shop or you are able to handle the bugs without disruption, maybe this is not a big reason for an OS upgrade.
Would there be a benefit to the company from more user-friendly features? There probably is no upgrade offering these days that does not boast of more user-friendly features and a host of wizards that can help the business accomplish its goals more easily. From V5R2 to V5R4, for example, the iSeries Navigator, the GUI Admin controls for the integrated Apache Web server, iSeries Access for the Web, HATS, WebFacing, and other modern capabilities have matured and stabilized and they have increased their range of facility. These user-friendly features can help you or your staff get started, learn new skills, answer questions, or offer easier ways to do things. These features are even more important to the computer novice who is just getting started with new tasks or projects.
Can you afford to wait and see? Being current does not necessarily mean bringing in the newest functions and deploying them before somebody else finds the bugs. If there is no immediate and compelling need for the function available in a new version of the operating system, it is more prudent to wait for a few months until the second cumulative PTF pack is out before you perform your upgrade. But please note that I am not suggesting waiting four or five years. By waiting just a few months, you can find out from others whether the upgrade is clean, and you can find out whether it would be worth the effort.
When it comes to operating system software, the company needs to get real work done and it is best to take your time, do the research, and make sure it’s what the company needs and that there are no big surprises. Though it is good to have an enriched computing experience it is not good to crash the system in order to get it. Likewise it is not good to sit back and let the clock tick time away while needed function sits on the shelf in your shop or in the IBM distribution center.
You sound like a busy guy, Jeb. I would bet your management team appreciates your hard work on their behalf. However, just like the Navy pilot who, against orders, landed on the carrier with his bombs to save taxpayers money, sometimes the money saved is not worth the risk and it is not worth what you have to give up to gain it.
I wish you the best.
Brian Kelly was an IBM Midrange SE for 30 years, and has spent nearly a decade as a System i5 consultant based in Scranton, Pennsylvania. He is also author of dozens of AS/400, iSeries, and System i5 books and an assistant professor at Marywood University, which uses the OS/400 and i5/OS platform and teaches courses in the box as well. Kelly is also one of the contributing technical authors to our Four Hundred Guru newsletter. He can be contacted through the IT Jungle contact page.