Purchasing Enterprise Examiner

  • QJo 2013-02-07 08:07
    Another wee problem ...
  • snoofle 2013-02-07 08:07
    ...because double manual entry without reconciliation fixes more problems than it creates?
  • Matt Westwood 2013-02-07 08:08
    Well hurry up in there for fuck's sake, I'm desperate ... to see my reports ...
  • snoofle 2013-02-07 08:10
    Interestingly, our users only care about getting their reports. Yet they've never complained about all the broken data on which they're based.
  • George Fitch 2013-02-07 08:10
    Let me guess... the real WTF is Microsoft Access?
  • Raedwald 2013-02-07 08:11
    Integration in the database layer often means trying to reconcile keys with no natural relationship


    Sounds better if you imagine this being spoken like a voice over from Burn Notice.
  • danskal 2013-02-07 08:17
    It's not often that you see a system/problem that is crying out for a buzzword solution, but this web of satan is rolling around in a sweaty fever, screaming "GIVE ME SOME SOA.... please, please, just a little bit of that ESB"
  • RuBen 2013-02-07 08:23
    "So why did you use SQL2008?"

    "Well it was a tip from our auditor. He was helping out the president's sick daughter."
  • Sahir Siddiqui 2013-02-07 08:26
    Oh yes the well known double entry technique to keep our new systems in sync, when the old systems they replaced never had any issues staying synced. Why am I not surprised.
  • CleanCode 2013-02-07 08:36
    TRWTF is no mention of a database architect, or any technical people between a VP and the poor IT guy.
  • DogsBody 2013-02-07 08:48
    snoofle:
    Interestingly, our users only care about getting their reports. Yet they've never complained about all the broken data on which they're based.


    I've had users complain when data was fixed as it broke their workarounds!
  • Mathias 2013-02-07 08:50
    I would love to read a Confession article from Remy.
  • Tony 2013-02-07 08:55
    I've fonud a problem!
  • Pock Suppet 2013-02-07 09:02
    DogsBody:
    snoofle:
    Interestingly, our users only care about getting their reports. Yet they've never complained about all the broken data on which they're based.


    I've had users complain when data was fixed as it broke their workarounds!


    This. Give me the friggin report now! No need for accuracy! No one actually looks at those numbers in the first place! It's all about relative numbers! Just make the change I told you to!

    Right until we get a client who *is* able to count past 20 and notices the report stupidity. Then it's time for a meeting where IT gets called on the carpet to explain why we "broke" things. Usually accompanied by the words "It never used to work like that!" And everyone thinks I'm crazy for never deleting emails...
  • RFoxmich 2013-02-07 09:06
    SELECT * FROM PURCHASE_ORDER_STAGE_TAB WHERE VENDOR = 'frist';
  • I heard it on the modem like 2013-02-07 09:06
    Sounds like a case of the FKIA report going way too far. http://www.centerpointforleaders.org/newsletters/FKIAReport.pdf

    CAPTCHA: nulla - When the moon hits the eye of a coder in the sky, that's a nulla.
  • Nagesh 2013-02-07 09:08
    The VP looks like he had some sense.
  • Confessor 2013-02-07 09:14
    I've been party to a system like this. Some Pointy-haired-boss (or customer) gets a wild idea that they want feature XYZZY or whatever buzzword is in vouge right now. Sit down and lay out a plan for the development strategy, requirements, development effort to develop said feature. Get told that's too much time by 2 orders of magnitude. Get handed some half baked piece of software (along with half baked expamples) and get told to integrate the solution. Still takes over half the time to get a moderately working solution. Get told to ship what we have (even though it's not done) with promises that we'll come back to finish it in a few months, but never do.

    CAPTCHA: I vink therefore i ven i am
  • renewest 2013-02-07 09:23
    Lost in translation...
  • John 2013-02-07 09:26
    "BizTalk doesn’t have a native connector for SQL2000 databases"

    Incorrect statement
  • RobFreundlich 2013-02-07 09:26
    Raedwald:
    Integration in the database layer often means trying to reconcile keys with no natural relationship

    Sounds better if you imagine this being spoken like a voice over from Burn Notice.

    Whaddya think, Mikey? Send Fi and Jesse to take a closer look at Purchize while you and I get some answers from the Oracle admin? Easy Peasy! But let's make it snappy - I've got plans with my ladyfriend this evening.
  • emaNrouY-Here 2013-02-07 09:35
    RFoxmich:
    SELECT * FROM PURCHASE_ORDER_STAGE_TAB WHERE VENDOR = 'frist';


    I found a bug... let me fix that for you!
    SELECT * FROM PURCHASE_ORDER_STAGE_TAB; WHERE VENDOR = 'frist';
  • Anonymous Coward 2013-02-07 09:38
    And I need remember to wear an athletic cup if I ever meet him.
  • Richard 2013-02-07 09:39
    Sounds like a good fix to me, actually. If the new system is causing a double-entry situation then it should be fixed - but that's far FAR better than having a poorly written fix that causes production data to be deleted. Not so WTF. I mean the environment is, but the VP made exactly the right call.
  • Swedish tard 2013-02-07 09:45
    emaNrouY-Here:
    RFoxmich:
    SELECT * FROM PURCHASE_ORDER_STAGE_TAB WHERE VENDOR = 'frist';


    I found a bug... let me fix that for you!
    SELECT * FROM PURCHASE_ORDER_STAGE_TAB; WHERE VENDOR = 'frist';


    And I've got a fix too...
    STAB STAB STAB STAB_STAB_STAB_STAB STAB STAB 'stab';
  • golddog 2013-02-07 09:51
    CleanCode:
    TRWTF is no mention of a database architect, or any technical people between a VP and the poor IT guy.


    Let's not suggest that is necessarily the solution.

    I overheard our recently-hired DBA/Data Architect talking with someone, explaining how to generate a script out of a database on SQL Server.

    "When you save the script, make sure to save it as ANSI, not Unicode. When you save as Unicode and check it into a repository, that messes up the file."

    I remain unimpressed.

    While I don't like the Captcha repeating meme, I find it sadly ironic that it's "genitus" for this post. Which might describe our DBA: not quite a genius.
  • Rich L. 2013-02-07 10:23
    Definitely one of my favourite WTF's of all time (up there with my friend the Shredder).

    Thanks for sharing.
  • DrPepper 2013-02-07 10:34
    [So, all of these other systems stopped merely dumping data into the Examiner, and started extracting data too.]

    An ad-hoc reporting system morphs into a "source of truth". There should have been about a hundred DBAs who choked on that idea; their role is the protection of the integrity of their data at all costs.

    Allowing data to flow back into one of their systems from an untrusted source should have raised red flags up the chain. And should have resulted in the firing of the people who allowed the data to get so screwed up.
  • Dave 2013-02-07 10:38
    The whole time I was expecting the article to end with some moral about not having P.E.E. (Purchasing Enterprise Examiner) in your pool (of data solutions).

  • HowItWorks 2013-02-07 10:56
    Dave:
    The whole time I was expecting the article to end with some moral about not having P.E.E. (Purchasing Enterprise Examiner) in your pool (of data solutions).

    TRWTF appears to be refering to Examiner, rather than using the acronym.
  • Another Anon 2013-02-07 10:59
    Any chance of trying to reorganize Payroll through this Oracle system? At least trial it out with the numb-nuts' own account information?
  • miker 2013-02-07 11:20
    [quote user="golddog"][quote user="CleanCode"]
    "When you save the script, make sure to save it as ANSI, not Unicode. When you save as Unicode and check it into a repository, that messes up the file."

    I remain unimpressed.[/quote]

    Really good advice, assuming he is not referring to UTF-8.

    A lot of tools will treat UTF-16, or UCS-2, files as binary. I can't remember which one SQL Server spits out by default, they are both bad UCS-2 is worse.

    I use an export tool that converts everything to UTF-8, but ASCII would be good enough for me.
  • Nagesh 2013-02-07 11:33
    golddog:
    CleanCode:
    TRWTF is no mention of a database architect, or any technical people between a VP and the poor IT guy.


    Let's not suggest that is necessarily the solution.

    I overheard our recently-hired DBA/Data Architect talking with someone, explaining how to generate a script out of a database on SQL Server.

    "When you save the script, make sure to save it as ANSI, not Unicode. When you save as Unicode and check it into a repository, that messes up the file."

    I remain unimpressed.

    While I don't like the Captcha repeating meme, I find it sadly ironic that it's "genitus" for this post. Which might describe our DBA: not quite a genius.


    Windows not support unicode complete like unix and linux.
    I could find severe links to this statement, but I can't bother to google right now.

  • C-Derb 2013-02-07 11:43
    Tony:
    I've fonud a problem!

    was it this?
    remy:
    He traced the tendrils of data back to the other systems, and fonud nothing amiss

    this?
    remy:
    So we use SQL2008 to proxy your data and pump records too and from Purchize.

    or this?
    remy:
    ... told the users that they simply had to keep manually enter their changes in both Purchize and Oracle.
  • Foo 2013-02-07 11:48
    snoofle:
    ...because double manual entry without reconciliation fixes more problems than it creates?


    without reconciliation, you'll never know what problems it creates, so yes.
  • snoofle 2013-02-07 11:53
    Dave:
    The whole time I was expecting the article to end with some moral about not having P.E.E. (Purchasing Enterprise Examiner) in your pool (of data solutions).

    A pool of PEE? This is becoming a theme!
  • cellocgw 2013-02-07 11:54
    <quote>Managing a long supply chain involves keeping thousands of moving parts in lock-step. From purchasing to order management, demand planning to operations, even the smallest hiccup in the chain can have massive impacts.</quote>
    There's the WTF in a nutshell. If your production system is not fault-tolerant, you're doomed no matter what.
  • cellocgw 2013-02-07 11:58
    Raedwald:
    Integration in the database layer often means trying to reconcile keys with no natural relationship


    Sounds better if you imagine this being spoken like a voice over from Burn Notice.


    +1 for that. I'd give you more points, but apparently that's frowned upon in these Web2.x days.
  • UK Pete 2013-02-07 12:05
    Pock Suppet:
    DogsBody:
    snoofle:
    Interestingly, our users only care about getting their reports. Yet they've never complained about all the broken data on which they're based.


    I've had users complain when data was fixed as it broke their workarounds!


    This. Give me the friggin report now! No need for accuracy! No one actually looks at those numbers in the first place! It's all about relative numbers! Just make the change I told you to!

    Right until we get a client who *is* able to count past 20 and notices the report stupidity. Then it's time for a meeting where IT gets called on the carpet to explain why we "broke" things.


    This wasn't a UK bank by any chance?
  • Lerch98 2013-02-07 12:20
    The technical term for this is Cluster Fuck
  • Shoreline 2013-02-07 12:28
    Pock Suppet:
    DogsBody:
    snoofle:
    Interestingly, our users only care about getting their reports. Yet they've never complained about all the broken data on which they're based.


    I've had users complain when data was fixed as it broke their workarounds!


    This. Give me the friggin report now! No need for accuracy! No one actually looks at those numbers in the first place! It's all about relative numbers! Just make the change I told you to!

    Right until we get a client who *is* able to count past 20 and notices the report stupidity. Then it's time for a meeting where IT gets called on the carpet to explain why we "broke" things. Usually accompanied by the words "It never used to work like that!" And everyone thinks I'm crazy for never deleting emails...


    Same.
    "How come you never delete emails?"
    "Leverage."

    A colleague of mine who wrote the reports was told one of his reports was "always wrong" based on the fact that two completely different numbers in the report were different.
  • chubertdev 2013-02-07 12:32
    Remy, time to update your schedule:

    1) caffeine
    2) author article
    3) more caffeine
    4) proofread article
    5) ?
    6) profit
  • James 2013-02-07 12:38
    The True WTF is the usual bizTalk solutions I see deployed. I am amazed at the number of folks out there who really think a COTS middle-ware layer is going to solve their problems.

    I suppose its a nice short cut for some situations; but really message routing, and workflow is the easy part of most middle-ware efforts. If you don't have a decent bench of systems people and developers its gonna be FAIL.

    Add to the problem that for some reason almost everyone who has ever built an orchestration in BizTalk thinks this somehow makes them Gods gift to the field systems integration. Weird culture issue there. Especially when they all run around designing things like in today's article. Seriously using a SQL2008 instance as proxy server? Really? Someone decided to go to production with that?

    Honest on the best BizTalk project outcomes I have saw was an organization where after a year long attempt to get their Commerce Server based website to put orders into their GL and fulfillment systems successfully, wound up with this:

    BT ( using the same database engine the site was on incidentally ) would pickup XML orders written out by the site to a directory on a file system, it would format them as more human targeted text then e-mail them to a customer service rep. The customer service rep would then manually enter the order, into both the accounting/GL systems and various fulfillment suff; which includes some sort of technician scheduling application, and creating receiving and picking tickets for the warehouse staff. Then if everything was okay, credit card accepted (yes the numbers were mailed to the CSR, but TLS was used so its okay right? ) the CSR would reply to the Biztalk e-mail adding the PO number to the subject. Biztalk would then do a few database look-ups get the final shipping cost ( rather than the badly estimated on made on line at order time), and build a recipient/confirmation all as a giant XML blob and drop it off in a file folder for Commerce Server to read at some point and wait for it: format it as a more human target message and mail it as the order confirmation....
  • Evan 2013-02-07 12:49
    Nagesh:
    Windows not support unicode complete like unix and linux.
    I could find severe links to this statement, but I can't bother to google right now.
    Hahahahahahahahaha
  • Michael Bolton 2013-02-07 12:54
    This has been my job for the last 7 years.
    Yee ha.
  • john 2013-02-07 13:05
    Better than the solution I expected to hear from the VP: "just use a shared spreadsheet".
  • Remy Porter 2013-02-07 13:23
    I actually added fonud to my spellchecker, just to toy with you all.
  • iMortalitySX 2013-02-07 13:53
    I think there is always a WTF if you draw a chicken and call it a data flow. Seriously though, am I the only one that sees a chicken in the drawing?

    CAPTCHA: vindico; Latin for avenge, punish, liberate, deliver, protect. I have seen chickens do none of these.
  • J 2013-02-07 13:56
    ...and told the users that they simply had to keep manually enter their changes in both Purchize and Oracle.

    Wanted: Data Entry Clerk
    Requirements: PL/SQL programming experience
  • Michael 2013-02-07 13:58
    This is simply the status-quo most places I have worked. I agree that it's broken and messed up, but hardly a WTF.
  • Kent 2013-02-07 14:04
    told the users that they simply had to keep manually enter their changes in both Purchize and Oracle.
    No biggie, give me a few hours and I can cobble up a new database for the original entry, which will feed both Purchize and Oracle.

    Solved!!

    ?
  • chubertdev 2013-02-07 14:07
    Remy Porter:
    I actually added fonud to my spellchecker, just to toy with you all.


    so are you going to award a prize the next time that this word is fonud in an article, and the commenter references this article?
  • DMJ 2013-02-07 14:08
    Somehow I missed the part where he put his resume out because this job was ... well, less than desirable.
  • Frank Lee 2013-02-07 14:16
    Then the auditor walked into the room, took one glance at the whiteboard, and had a heart attack then and there.

    They rushed him to the hospital where, unfortunately, the president's daughter and the grad student were being treated for a highly virulent condition.

    When the auditor realized he was going to die, he called the VP of Global Purchasing, not wanting the monstrosity he had found to live on in his absence. (That's the whiteboard monstrosity, not the president's monstrosity, in case you're wondering.)

    Unfortunately the highly virulent condition was the first virus that has both physical and electronic components, so it spread over the phone line and killed the VP of Global Purchasing.

    I love a happy ending!
  • Matt Westwood 2013-02-07 14:39
    Dave:
    The whole time I was expecting the article to end with some moral about not having P.E.E. (Purchasing Enterprise Examiner) in your pool (of data solutions).



    Just occurred to me that yesterday's auditor ought to have asked questions about memory leaks ...
  • Chuck Finley 2013-02-07 14:46
    Raedwald:
    Integration in the database layer often means trying to reconcile keys with no natural relationship


    Sounds better if you imagine this being spoken like a voice over from Burn Notice.


    I agree!
  • Some Damn Yank 2013-02-07 14:55
    Pock Suppet:
    Then it's time for a meeting where IT gets called on the carpet to explain why we "broke" things.
    I worked on a project once involving migration to newer workstations/OS versions. The team for one in-house program said they could not guarantee it would give accurate results on the new platforms, only that it would give the same results as it gave on the old platforms. I found that position refreshingly honest and pre-emptively butt-covering.
  • Some Damn Yank 2013-02-07 15:08
    iMortalitySX:
    I think there is always a WTF if you draw a chicken and call it a data flow. Seriously though, am I the only one that sees a chicken in the drawing?
    Yes, you are.
  • wldkrd1 2013-02-07 15:40
    Raedwald:
    Integration in the database layer often means trying to reconcile keys with no natural relationship


    Sounds better if you imagine this being spoken like a voice over from Burn Notice.


    I prefer a voice over from March of the Penguins
  • Mitch 2013-02-07 15:45
    chubertdev:
    Remy, time to update your schedule:

    1) caffeine
    2) author article
    3) more caffeine
    4) proofread article
    5) toilet break
    6) profit

    FTFY
  • Jazz 2013-02-07 15:46
    TRWTF is that managers / executives still think allowing mission-critical systems to be built with this kind of a design is a good idea.

    Okay, but seriously. Take your average VP at a company that has a long supply chain managed with some enterprise software. Somewhere between birth and becoming employed as an executive, one presumes that the VP underwent a process of education, in which they transitioned from a state of not-knowing-how-to-manage to a state of knowing-how-to-manage. Right? This can be reasonably inferred from the fact that the VP was judged to be qualified for their job when they were hired. And part of that educational process involves becoming familiar with documented cases of poor management choices in the past, so that (ostensibly) the VP can avoid the mistakes that other executives have made. Right? And there are numerous documented cases of management choosing a system with a half-assed architecture like this because it's cheaper, and then later coming to regret it, over the past 15-20 years. Right? And, given that this education is in the field of management theory & practice, rather than in the field of software engineering, the many reasons to avoid this architecture are expressed (and learned) in business terminology (i.e. "negatively impacts revenue streams") that the students can understand, rather than in technical terminology (i.e. "fails validation and rolls back") that they might not.

    Therefore, anyone who is judged to be qualified for an executive position in today's market has already been educated on the dangers that this kind of system architecture poses for the company's bottom line and long-term health, and knows to avoid it.

    Therefore, any executive who continues to advocate in favor of such a system is knowingly risking the long-term prospects of the company, and therefore is poorly-qualified for their job. And this should be reflected on their reviews (and if it's not, their reviewer is poorly-qualified to be reviewing them).

    (Captcha: 'minim' -- the required amount of intelligence to land a $400k job.)
  • juan 2013-02-07 15:54
    wldkrd1:
    Raedwald:
    Integration in the database layer often means trying to reconcile keys with no natural relationship


    Sounds better if you imagine this being spoken like a voice over from Burn Notice.


    I prefer a voice over from March of the Penguins
    Integration in the database layer often means trying to reconcile keys with no natural relationship....These are their stories.....
  • Z 2013-02-07 16:36
    Michael:
    This is simply the status-quo most places I have worked. I agree that it's broken and messed up, but hardly a WTF.


    TRWTF is the fact that you agree that it is broken and messed up, but don't see the WTF.
  • chubertdev 2013-02-07 16:53
    Mitch:
    FTFY


    Ahhh, you ruined it.
  • PedanticCurmudgeon 2013-02-07 16:53
    golddog:
    I overheard our recently-hired DBA/Data Architect talking with someone, explaining how to generate a script out of a database on SQL Server.

    "When you save the script, make sure to save it as ANSI, not Unicode. When you save as Unicode and check it into a repository, that messes up the file."
    Please tell me you're not still using Source Safe.
  • Soviut 2013-02-07 16:54
    Shoreline:
    Pock Suppet:
    DogsBody:
    snoofle:
    Interestingly, our users only care about getting their reports. Yet they've never complained about all the broken data on which they're based.


    I've had users complain when data was fixed as it broke their workarounds!


    This. Give me the friggin report now! No need for accuracy! No one actually looks at those numbers in the first place! It's all about relative numbers! Just make the change I told you to!

    Right until we get a client who *is* able to count past 20 and notices the report stupidity. Then it's time for a meeting where IT gets called on the carpet to explain why we "broke" things. Usually accompanied by the words "It never used to work like that!" And everyone thinks I'm crazy for never deleting emails...


    Same.
    "How come you never delete emails?"
    "Leverage."


    People still delete emails? With inboxes in the tens of gigabytes these days, you'd have to be an awfully compulsive person to bother hitting delete instead of "read next".
  • PedanticCurmudgeon 2013-02-07 16:55
    Jazz:
    TRWTF is that managers / executives still think allowing mission-critical systems to be built with this kind of a design is a good idea.

    Okay, but seriously. Take your average VP at a company that has a long supply chain managed with some enterprise software. Somewhere between birth and becoming employed as an executive, one presumes that the VP underwent a process of education, in which they transitioned from a state of not-knowing-how-to-manage to a state of knowing-how-to-manage. Right? This can be reasonably inferred from the fact that the VP was judged to be qualified for their job when they were hired. And part of that educational process involves becoming familiar with documented cases of poor management choices in the past, so that (ostensibly) the VP can avoid the mistakes that other executives have made. Right? And there are numerous documented cases of management choosing a system with a half-assed architecture like this because it's cheaper, and then later coming to regret it, over the past 15-20 years. Right? And, given that this education is in the field of management theory & practice, rather than in the field of software engineering, the many reasons to avoid this architecture are expressed (and learned) in business terminology (i.e. "negatively impacts revenue streams") that the students can understand, rather than in technical terminology (i.e. "fails validation and rolls back") that they might not.

    Therefore, anyone who is judged to be qualified for an executive position in today's market has already been educated on the dangers that this kind of system architecture poses for the company's bottom line and long-term health, and knows to avoid it.

    Therefore, any executive who continues to advocate in favor of such a system is knowingly risking the long-term prospects of the company, and therefore is poorly-qualified for their job. And this should be reflected on their reviews (and if it's not, their reviewer is poorly-qualified to be reviewing them).

    (Captcha: 'minim' -- the required amount of intelligence to land a $400k job.)
    7/10 would flame if in bad mood (sorry, can't do "Not Bad" meme from work)
  • Your Name 2013-02-07 17:00
    Jazz:
    Take your average VP at a company that has a long supply chain managed with some enterprise software. Somewhere between birth and becoming employed as an executive, one presumes that the VP underwent a process of education, in which they transitioned from a state of not-knowing-how-to-manage to a state of knowing-how-to-manage. Right? This can be reasonably inferred from the fact that the VP was judged to be qualified for their job when they were hired.


    I'm gonna stop you right there. Your average VP, particularly in a WTF-heavy company like the ones that show up here, was hired into the position because they know a guy who knows a guy who knows a guy, so to speak. Now, I'm not saying every VP is like that, but there's a large percentage of them that are, and they're more likely to end up on this site.
  • Nim 2013-02-07 17:21
    I think I stumbled upon something similar in a D&D adventure some time ago. But there it was called the Demonweb Pits.
  • Simon 2013-02-07 17:33
    Good call from the VP. Entering data into two systems in parallel might be annoying and error prone, but as an immediate fix, it's absolutely the right thing to do. At worst, it's less of a disaster than the current system, and doesn't have the risk of trying to make changes to the wider system while under time pressure.

    Clued-up management... who'd have thought it possible?
  • Harrow 2013-02-07 18:27
    Kent:
    told the users that they simply had to keep manually enter their changes in both Purchize and Oracle.
    No biggie, give me a few hours and I can cobble up a new database for the original entry, which will feed both Purchize and Oracle.
    Good idea.

    Microsoft Access is an excellent tool for this, but Tony already has an Access application or two in his snake pit. For extended job security I would have to suggest Borland Paradox for Windows v. 7/32, which seems to be the only remotely db-related tool they are not already using.

    -Harrow.
  • Barfo Rama 2013-02-07 19:23
    You get change requests by email?
  • PiisAWheeL 2013-02-07 19:38
    Some Damn Yank:
    iMortalitySX:
    I think there is always a WTF if you draw a chicken and call it a data flow. Seriously though, am I the only one that sees a chicken in the drawing?
    Yes, you are.
    I see the chicken. I had to get really high first tho.
  • The Joker 2013-02-07 20:16
    It's simple, we ummm...

    We kill the batman.
  • Paul 2013-02-07 21:13
    Your Name:
    Your average VP, particularly in a WTF-heavy company like the ones that show up here, was hired into the position because they know a guy who knows a guy who knows a guy, so to speak.
    Let's just assume for discussion that this is true. The reasonable follow-up question is... Why?

    Take two companies in a similar line of business. One is run by a bunch of clueless pals. The other hires smart competent people. Wouldn't the good company kick the bad company's butt all over town?

    My theory: taxpayer funded bailouts to keep the stupid ones afloat?
  • The Crunger 2013-02-07 23:45
    Jazz:
    TRWTF is that managers / executives still think allowing mission-critical systems to be built with this kind of a design is a good idea.

    Jazz:

    Therefore, anyone who is judged to be qualified for an executive position in today's market has already been educated on the dangers that this kind of system architecture poses

    Which manager actually praised the system design? Most of the time, even the dimmer ones realize that a monster is growing, while they themselves bolt on one irregular chunk at a time.

    It's not that the knowledge of how to achieve a database schema, fully compliant with 3NF hurts, but using that knowledge really doesn't help.

    Your job is to make your SBU succeed -- nothing else. When the business systems your SBU needs aren't in place, it's time to pick, purchase, staff, and integrate. The best choice for your SBU is to add your our particular streaming contribution to "the pool". You can advocate for the greater good after you've dodged the reaper.
  • urza9814 2013-02-08 10:15
    Paul:
    Your Name:
    Your average VP, particularly in a WTF-heavy company like the ones that show up here, was hired into the position because they know a guy who knows a guy who knows a guy, so to speak.
    Let's just assume for discussion that this is true. The reasonable follow-up question is... Why?

    Take two companies in a similar line of business. One is run by a bunch of clueless pals. The other hires smart competent people. Wouldn't the good company kick the bad company's butt all over town?

    My theory: taxpayer funded bailouts to keep the stupid ones afloat?


    The less consumer-oriented a company is, the dumber it can be. If Amazon.com pulled this kinda crap, their service would start to suck and people would start switching to alternatives. Although they'd have enough momentum to survive quite some time...

    But this doesn't seem like the kind of company that deals directly with consumers. They get a few large contracts and they're good. And they get those contracts from the company owned by the CEO's golfing buddy...or, of course, from his uncle, Sam...
  • eVil 2013-02-08 10:23
    PiisAWheeL:
    Some Damn Yank:
    iMortalitySX:
    I think there is always a WTF if you draw a chicken and call it a data flow. Seriously though, am I the only one that sees a chicken in the drawing?
    Yes, you are.
    I see the chicken. I had to get really high first tho.


    TRWTF is anyone who didn't see the chicken.
  • ObiWayneKenobi 2013-02-08 10:32
    The Real WTF is how many managers/higher-ups actually think that yelling at people is a positive way to get anything accomplished. I thought we all learned in kindergarten (or maybe first grade) that screaming and throwing a tantrum isn't how you deal with people.
  • Mike 2013-02-08 10:55
    that is why it is essential to have covering fire. Preferably with heavy rifles and rocket launchers.
  • VJG 2013-02-08 13:26
    Yes, I see the chicken. I was going to announce it and thought I'd better see if anyone else had already pointed it out.

    Now I can't finish reading the article. I keep seeing that blasted chicken in my peripheral vision.

    It frightens me.
  • D-Coder 2013-02-08 13:50
    Matt Westwood:
    Dave:
    The whole time I was expecting the article to end with some moral about not having P.E.E. (Purchasing Enterprise Examiner) in your pool (of data solutions).



    Just occurred to me that yesterday's auditor ought to have asked questions about memory leaks ...
    Time to tell the VP "You're in over your head."
  • Nagesh 2013-02-08 14:05
    Soviut:

    People still delete emails? With inboxes in the tens of gigabytes these days, you'd have to be an awfully compulsive person to bother hitting delete instead of "read next".

    Only 200 MB allocation from Microsoft Exchange here.
  • justme 2013-02-08 16:56
    golddog:
    CleanCode:
    TRWTF is no mention of a database architect, or any technical people between a VP and the poor IT guy.


    Let's not suggest that is necessarily the solution.

    I overheard our recently-hired DBA/Data Architect talking with someone, explaining how to generate a script out of a database on SQL Server.

    "When you save the script, make sure to save it as ANSI, not Unicode. When you save as Unicode and check it into a repository, that messes up the file."

    I remain unimpressed.

    While I don't like the Captcha repeating meme, I find it sadly ironic that it's "genitus" for this post. Which might describe our DBA: not quite a genius.


    Actually, they are partly right , but probably did not explain it correctly. I have run into Unicode/ANSI issues when using VS and MSSM and the export feature.
  • justme 2013-02-08 17:14
    DrPepper:
    [So, all of these other systems stopped merely dumping data into the Examiner, and started extracting data too.]

    An ad-hoc reporting system morphs into a "source of truth". There should have been about a hundred DBAs who choked on that idea; [bold]their role is the protection of the integrity of their data at all costs. [/bold]

    Allowing data to flow back into one of their systems from an untrusted source should have raised red flags up the chain. And should have resulted in the firing of the people who allowed the data to get so screwed up.


    Okay, help needed. Not a formally trained DB but fell into it. I was arguing with someone the other day. Should users be allowed to delete records that they create ( that have not be used in any transaction ). I was told that it should stay, but we could transition it to the last state [final]. I objected because I saw this as 1) mucking up the state of "final". Final implies it has gone through all the states including released. 2) while general users would not be able to see he data, it would still be sitting there for the owner 3) it would skew metrics 4) why allow known junk to stay.

    Thoughts ?
  • justme 2013-02-08 17:28
    Jazz:
    TRWTF is that managers / executives still think allowing mission-critical systems to be built with this kind of a design is a good idea.

    Okay, but seriously. Take your average VP at a company that has a long supply chain managed with some enterprise software. Somewhere between birth and becoming employed as an executive, one presumes that the VP underwent a process of education, in which they transitioned from a state of not-knowing-how-to-manage to a state of knowing-how-to-manage. Right? This can be reasonably inferred from the fact that the VP was judged to be qualified for their job when they were hired. And part of that educational process involves becoming familiar with documented cases of poor management choices in the past, so that (ostensibly) the VP can avoid the mistakes that other executives have made. Right? And there are numerous documented cases of management choosing a system with a half-assed architecture like this because it's cheaper, and then later coming to regret it, over the past 15-20 years. Right? And, given that this education is in the field of management theory & practice, rather than in the field of software engineering, the many reasons to avoid this architecture are expressed (and learned) in business terminology (i.e. "negatively impacts revenue streams") that the students can understand, rather than in technical terminology (i.e. "fails validation and rolls back") that they might not.

    Therefore, anyone who is judged to be qualified for an executive position in today's market has already been educated on the dangers that this kind of system architecture poses for the company's bottom line and long-term health, and knows to avoid it.

    Therefore, any executive who continues to advocate in favor of such a system is knowingly risking the long-term prospects of the company, and therefore is poorly-qualified for their job. And this should be reflected on their reviews (and if it's not, their reviewer is poorly-qualified to be reviewing them).

    (Captcha: 'minim' -- the required amount of intelligence to land a $400k job.)


    You are underestimating the effects of hubris on executive decision making. They have heard the stories, but believe they are immune.
  • observer 2013-02-09 06:15
    Any company that implements a reporting ecosystem as haphazard as this deserves all the bad things it could get. The WTF is that the company did not sink.
  • TheCPUWizard 2013-02-09 21:21
    [quote user="justme\Okay, help needed. Not a formally trained DB but fell into it. I was arguing with someone the other day. Should users be allowed to delete records that they create ( that have not be used in any transaction ). I was told that it should stay, but we could transition it to the last state [final]. I objected because I saw this as 1) mucking up the state of "final". Final implies it has gone through all the states including released. 2) while general users would not be able to see he data, it would still be sitting there for the owner 3) it would skew metrics 4) why allow known junk to stay.

    Thoughts ? [/quote]

    Two parts:

    1) Data should be validated and verified before hitting the "main system"...there should be NO junk...

    2) Once data exists it is part of an audit trail and must be subject to approved formal retention policies (many of which are legally mandated)..in general this means no deletion...using 'final' may be a WTF, and an appropriate solution *might* be moving to some other datastore...but having the data "simply be gone" (bfore expiration according to retention policy] is a major WTF.

    (note that if this was followed in the story, thre would have been no story....]
  • Rob 2013-02-10 14:56
    Sounds to me like that corporation misses the concept of data ownership. Different corporate systems almost always end storing similar kinds of data (like product information) and also end up publishing data to other systems. There is nothing really wrong with this, but errors occur when a local system fail to identify who owns which records. The local system can delete those records that it owns and notify other systems that the local system has deleted that record. A local system can delete records that it does not own(remove them from its system) , but the local system can't delete records owned by another system and tell other systems to remove them as well.

    The same thing applies to updates. If a local system owns the record, then only it can change the record and publish the change, but when the local system does not own the record, it can only update the record within the scope of the local system.

  • nasch 2013-02-24 16:31
    Jazz:
    TRWTF is that managers / executives still think allowing mission-critical systems to be built with this kind of a design is a good idea.


    I don't think "design" is really the right word here.