Unconfigured Ad Widget

Collapse

Announcement

Collapse
No announcement yet.

Happy New Year (Sort Of)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Happy New Year (Sort Of)

    Ralph, I read an article about the computer system Walmart uses. The more I read, the more I kept saying to myself "that sounds like an AS/400". The article stated it was a Teradata system, which isn't something I have ever heard of. They are totally centralized, except for their web presence, which is externalized. One huge multi-processor machine managing all of it. All the data feeds back to HQ on a near realtime basis. As apposed to the way most retail chains do it by sending batches of data to corporate HQ on daily or weekly reportint schedules. Most everything Walmart has (software wise) was custom built.

  • #2
    Happy New Year (Sort Of)

    Hi Nettie! I don't understand your statement. You can still run programs originally written on a System/3 on today's iSeries boxes. I can't name one other platform for which this is true. You can take a compiled program object and move it from one release to another with little or no difficulty. Find me a Windows 3.1 program of any complexity that will still run on Windows XP (and the ones that run on XP won't run on Vista). Drivers have to be rewritten for every release, and I still can't share a printer between my Windows 2003 Server and my Windows XP desktop without jumping through hoops. On the iSeries, my application can be copied to a different machine which runs on a completely different processor, and the program never knows it. I'm sorry, but I just don't understand your point. Joe

    Comment


    • #3
      Happy New Year (Sort Of)

      Hi Shane, Yes, from reading in the past that's exactly the impression I had, they were making good use of AS/400's. I don't have anything at the tip of my fingers saying so, but that was my impression. Teradata is a very good data warehouse from NCR. Frequently discussions about the value of the AS/400 are shunted off into data warehouse and BI talk, point and click analysis querying, etc. The people writing may not even know the difference between transaction production work and data warehouse data mining, I don't know. In any event, everything I write about RPG and the AS/400 concerns the actual production and processing of data that can optionally end up in a data warehouse. While there are fundamental differences in production work, and whether it can work or not at all, as I described earlier, discussion of data warehouses is more of just a preference thing. They work very well, and in fact the government cites its data warehouses as a success story. The GAO points out to them that a data warehouse is not an FBI Case processing system, in other words they aren't fooled even if the FBI thinks they are fooling anybody. DB2/400 with IBM's bitmapping indexing can handle data warehousing very well in my opinion, there is no need for something like Teradata if you are an AS/400 shop. But it is popular and Teradata people know nothing of the AS/400 and want to know nothing of it. Basically they say FTP files to us or send transactions on a message queue product and go away. Queries are then done with a Windows SQL query product but of course they all work against the AS/400 as well. A very large retailer I worked for runs on AS/400's but put their data warehouse on Teradata. It all could have been done on an AS/400, in fact the reporting was compared against what we had on the AS/400 to validate it before putting it into production, but again, Teradata is a fine system and widely used in data warehousing. Transaction work on the AS/400 and data mining on Teradata would be analogous to what I write above for the government as well. Glad you thought the same thing about Wal-Mart, Shane. rd

      Comment


      • #4
        Happy New Year (Sort Of)

        Ralph, You said: DB2/400 with IBM's bitmapping indexing... What does this mean? Are you referring to EVI's, etc? If you start talking about DB2/400 binary radix trees next, I'll just assume you mean a LF. Chris

        Comment


        • #5
          Happy New Year (Sort Of)

          Yes, EVI's. Bitmapped indexes which are ideal for data mining, Chris. rd

          Comment


          • #6
            Happy New Year (Sort Of)

            Hi Ralph, I find it humorous when you invent your own technical term. And I think you lose a little credibility too. Do a Google on "BITMAPPING INDEXING" and see you many hits you get. EVI is the proper term for the iSeries. I'm aware of what EVI's are when it's the right tool for the job. I wouldn't create one over a timestamp, for example. Regards, Chris

            Comment


            • #7
              Happy New Year (Sort Of)

              Actually, encoded vector index (EVI) is IBM's acronym for a bitmapped index (also called a bitmap index). The concept has been around since at least the 90's, and is pretty widely used. In the index, each field is represented by a bitmap where each bit stands for one possible value of the field. With this technique, you can easily create queries checking for one or more specific values by creating a mask for the required bits and then AND'ing it with the value from the record. Google for bitmap index or bitampping index. Then you might want to rethink your comment to Ralph. Joe

              Comment


              • #8
                Happy New Year (Sort Of)

                Thanks for the clarification, Joe. Yes, Chris, they are not good for our production work with unique keys and master-detail relationships, etc. But data mining is a different thing altogether. You're probably familiar with those stories of strange patterns found in the retail industry, like guys purchasing beer and diapers at 5 pm. (at least I think that was it. In any event, data mining is often over fields we wouldn't index with a logical. I believe IBM explains EVI as ideal for many records with just a few values, the opposite of unique. The results are very fast for data warehouse pattern search type queries, which include the timestamp, as we saw from the example above. Cheers Joe and Chris. rd

                Comment


                • #9
                  Happy New Year (Sort Of)

                  Hi Joe, Google for bitmap index or bitmapping index. Ralph specifically said "bitmapping indexing". To me, this is like prounouncing the name of wine a incorrectly, makes you sound silly. Maybe I'm being too picky. And, we're talking about the iSeries here (Ralph said "DB2/400"). The term EVI should have been used in the first place. In my meetings, I'm going to start saying "binary radix tree" instead of "logical file" and see if I get any strange looks. Do you think I will? Have a good one! Chris

                  Comment


                  • #10
                    Happy New Year (Sort Of)

                    Well, if you want to be picky, you didn't say he pronounced it incorrectly, you said he made it up. To me, that made it sound like you didn't think there was any such term, and I just wanted to clear that up. The rest of the world calls an EVI a bitmap(ped) index. As to using the correct "iSeries" name, I understand your point. I still get a little nutty when people talk about DDS defined files and using the terms "tables", "rows" and "columns". But live and let live. Joe

                    Comment


                    • #11
                      Happy New Year (Sort Of)

                      Chris, if your meeting is with people outside of the AS/400 world, which is what my letter to InformationWeek was and followup posts within the same context, then neither logical file nor EVI will get you very far. I also will surely mispronounce any given wine you have in mind. I will drink it though. rd

                      Comment


                      • #12
                        Happy New Year (Sort Of)

                        Ralph specifically said "bitmapping indexing". To me, this is like prounouncing the name of wine a incorrectly, makes you sound silly. I also specifically said "bitmapped indexes" when you asked for a clarification. rd

                        Comment


                        • #13
                          Happy New Year (Sort Of)

                          In 2004, prompted by no action at all to promote its iSeries, I whipped up thirty-nine chapters of concern and advice to Big Blue in the form of a book titled, Can the AS/400 Survive IBM? Over the past year and a half since writing this call to action, I have been somewhat quiet about IBM being the main reason (past, present, and future) for my favorite product's lackluster performance in the marketplace. Quite frankly, though way too slow, I do see some iSeries movement in IBM with the now not so "new" iSeries management team. However, like you, I have yet to see anything earth shattering or overly innovative. There has been no real marketing or advertising campaign that has reached my neighbors, for example, and without our neighbors knowing this box exists, it is destined for continued obscurity. Moreover, the machine still does not support Web applications in RPG, the major platform language; it is still not delivered with a modern GUI interface; it is not even mentioned in major College and University textbooks so students do not have a natural way of knowing that the box even exists, and if it were not for its strong following, it would probably already be gone. Ironically, the parts of the iSeries that have been highlighted with the i5, though high in innovation and admirable technology, are not the parts of an iSeries of most interest to the small business shops who use the platform. I think it is wonderful that the iSeries can run all this stuff and even at the same time, but it has nothing to do with why I recommend iSeries to my clients. By the way, doesnít all that other stuff run on other IBM boxes and non-IBM boxes that don't cost half as much as the iSeries? My clients do not buy an iSeries to run Linux, AIX, or for firewall partitioning. They buy an i5 because it comes with the most advanced business-oriented operating system in the world -- OS/400 (i5/OS) and because it comes with the finest business language in the world (all flavors of RPG) and of course because it has an integrated database, major fault-tolerance, the CL language, display files, and tons of applications that work day-in and day-out with no issues. The real irony is that these parts that keep the platform alive in the minds of its constituents are the parts that IBM continues to de-emphasize in favor of the glitz and the glamour of high technology. Yet, with iSeries, the glitz and glamour brings in few additional customers. The magic of OS/400 and i5/OS and a great companion business language, such as RPG or as I think it should be called, the International Business Language, has always been the essence of the iSeries platform. It still is. A few hundred thousand total installed units in the world, growing at the torrid pace of perhaps 60,000 per good year, against Intelís zillions is not going to create or increase iSeries mindshare anytime soon. That will happen only if IBM management thinks that an expanded mindshare is necessary to IBM managementís success. Increasing mindshare has some risks that many in IBM continue to be unwilling to take. Yet, most of us out here where the rubber meets the road know in our guts that the biggest hope for the platform is if other people than us can find out about it. However, unless iSeries management decides that the iSeries is to become a pervasive machine, other people are not ever going to hear about it in any meaningful way. IBMís big fear among other fears is that it would have to position the iSeries to compete against other servers such as those running Windows and Unix, (AIX and Linux) and its coopetition partners may not like that. Though IBM will take whatever iSeries business comes its way, the company is not inclined to declare war on Unix, Linux, and/or Windows. Yet, you and I, the moon, and even Bart Simpson know that is the only way the iSeries can win mindshare and market share. As long as IBM sells Intel servers, however, that cannot happen. IBM just does not have the guts to do it. Ask yourself when the last time was that you heard or saw an iSeries Advertisement. You know IBM believes in advertising because you have seen lots of xSeries ads with their Intel processors. Marketing exposure however, is not all that is needed. At the server level, IBM has not positioned the i5 to sell exceptionally well even if it were advertised well. It needs to be more affordable and (now don't laugh) it would also help to have a nice inexpensive GUI-driven i5/OS desktop version with mini-server facilities and whatever else is necessary to make it successful. Why is it so expensive? Because only 60,000 are sold each year, thatís why! Keeping the machine a secret of course is not going to increase sales. It is certainly not the Power processors that are keeping the iSeries price high. When Microsoft can pack almost 10,000 CPW of power on three IBM Power chips in an under-$500.00 game console called Xbox 360, what price do you think IBM should be charging for 30 CPW of interactive on an i5 model 520? IBM takes Intelís market development dollars and in its TV ads, the company never fails to mention that the IBM eServer xSeries is powered by those phenomenal Intel processors. Why would anybody believe that IBM has a better processor? Yet, IBMís best scientists from Rochester to East Fishkill and beyond will tell you privately how great the Power platform is compared to Intel wares. IBM is so fearful about risking the loss of those dollars (bribe money) that Intel gives it to push the Intel processor line, Big Blue is not about to upset its biggest competitor Ė Intel. The line of least resistance has been for IBM to promote and appease Intel while promoting its own eServer xSeries. If IBM were to promote the i5, somebody in IBM not Intel, would have to fund the campaign Ė and can you imagine how upset that would make the mighty Intel? I thank Bob Cozzi and all of those of you who speak to IBM to help the Rochester contingency make the iSeries the success that it should be and once was. If IBM thinks we are all pleased or we are complacent with its iSeries efforts, chances are that nothing good or innovative will happen to the iSeries. Yet, IBM's new iSeries management team absolutely wants to be successful. Their careers are riding on it. So, it is up to us -- all of us, to let IBM know when we are not pleased with its efforts. IBM needs to know where we want them to take this marvelous all-everything machine. One of my big iSeries fears is that a small growth in iSeries sales may make the current management team successful enough in IBM that they do not have to do anything radically strategic, such as reducing price, making the box easier to obtain, and making the platform ideal for mom and pop ventures. The good news is that we will not have to wait long to know. I think the answers regarding IBMís platform commitment will be out there within the next six months. By then, we will know whether we have been witnessing lip service or the real thing. Within six months, I hope to see the product on a clear course that makes it fly high or I expect to know and you will know that it is not IBM's desire to do so. I noticed that a number of iSeries professionals chose to comment on this article. I would encourage many more. Please do not leave it to chance. Let you voice be heard so loud that the easiest thing for IBM to do is to appease you and I instead of Intel. While you are at it, you can get some good fodder for material all across the Internet. I found this 2002 article on IBMís Sam Palmisano very telling about the culture that Big Blue must overcome to be innovative. I would recommend your checking it out. ďFixing IBM Big Blue has big problemsĒ Robert X. Cringely http://www.pbs.org/cringely/pulpit/pulpit20020418.html To keep IBM management honest to its constituencies, for my money, I would like to see the company invite brave analysts like Bob Cozzi, Timothy Prickett Morgan, Bob Tipton, Thomas Stockwell, along with some brave IT managers from small development shops to give the IBM chiefs the personal once-over. IBM can learn a lot from these folks. I would also like to see IBM respond point by point to each of the concerns that are raised by its industry analysts and by its loyal constituency. Wouldnít you like to read a point by point reply by IBM to Bob Cozziís article? Much of the legitimate complaining that IBM endures about its iSeries line comes about because iSeries customers really care and they do not have clear and irrefutable evidence that IBM cares as much. The question of whether the AS/400 can survive IBM should have been answered by IBM long ago. Yet, somehow, because IBM has not answered the question satisfactorily, iSeries customers are compelled to keep asking. And that is good. Thank you Bob Cozzi.

                          Comment


                          • #14
                            Happy New Year (Sort Of)

                            I think the Teradata hardware is NCR. At least that's what their website hardware points to.

                            Comment


                            • #15
                              Happy New Year (Sort Of)

                              Very nice post Brian. When you said "it is still not delivered with a modern GUI interface", this really hit home. We provide a complete turn key package to our customers. All they have to do is manage the AS/400 (eg. run backups, re-enable profiles, upgrade it, etc.). My customers complain that the AS/400 is too hard to use. They cannot remember the commands. One manager said to me "I've been doing this for 3 years and I still can't find my way around, but the Windows server is easy". That manager has decided to migrate off the AS/400. I can not fault this manager for his opinions. After all he doesnít use our package, the users do. His only contact with the machine is through this green screen command system. He admits that the machine never goes down and requires very little attention, but he wants something he can figure out and he already has several Windows servers. When I here these types of complaints, I always ask them if they have tried using OpsNav. Then my customers tell me either that it is slow or it doesn't work (of course I, as well as the rest of you already knew this because of past experiences). OpsNav however, isn't a GUI either, nor is it based on anything that can be used for applications so they can be integrated again. So it was a valiant attempt to eliminate the demand for a GUI, but it failed. In my opinion it is also a great demonstration of why Java wonít make it in the real world. One of my sticking points is the lack of a GUI. This system needs a built in GUI (supported by IBM!) that can be used to run apps and the operating system alike. We used to have a hardware accelerated UI that was totally in charge and fully integrated into the machine (eg. paging, long waits, time slice, priority, etc.). Most importantly 5250 was efficient and economical to develop and run. I say "used to have" because it is now relegated to the "legacy" pile. Rightfully so, but there needs to be a replacement first!! Before any of you even start to think itÖ.WEB pages are not a replacement. Not even close. WEB pages have their place, such as information retrieval. WEB pages running on HTML and the plethora of related languages required to make a web page run are no replacement for what we had. We used to be able to create tools with screens and commands and share them with each other and easily imbed in-house and packaged tools into our apps. I havenít seen a matching occurrence on the WEB side. That was part of the community of AS/400 programmers that has been lost. We are all embarking on different GUI development paths attempting to recreate what we had and in many cases using third party products to fill the void left by IBM. The RPG fragmentation has caused a similar divergence among the flock. But, language development will eventually wane and this group will congeal once more. What is left of them anyway. As many have already written, programmers are leaving for Windows platforms. I still hold out hope, that a browser based, easy deployment, centrally managed, light weight client could eventually emerge from IBM. This would be the answer we have all been waiting for. If something doesnít happen soon, it just wonít matter. Food and shelter are primal motivators. In the mean time .NET is looking very interesting. Especially the parts about, centrally managed, browser based, easy deployment and the elimination of DLL hell. Oh, and donít forget its GUI. The proof will be in the pudding. Havenít seen much .NET yet. If anyone has seen a .NET app or tool that runs from the AS/400 data base let me know. It would be much appreciated. Shane.

                              Comment

                              Working...
                              X