Slashdot Mirror
Ask Slashdot: Biggest IT Management Mistakes?
snydeq writes: Sure, nobody's perfect. But for those in charge of enterprise technology, the fallout from a strategic gaffe, bad hire, or weak spine can be disastrous, writes Dan Tynan, in an article on the biggest management mistakes in IT. "Some of the most common IT gaffes include becoming trapped in a relationship with a vendor you can't shake loose, hiring or promoting the wrong people, and hiding problems from top management until it's too late to recover." What are some other career- and company-destroyers you've witnessed in your years in IT?
Implementing SAP
Outsourcing
Outsourcing your SAP implementation
People in India can't possibly know your business, your other employees or your customers as well as qualified, competent, real live boots on the ground in the US of A.
The two biggest mistakes I see is that a dev team which is doing fine gets cut and the people outsourced or offshored. Sales and quality hit the shitter, but management doesn't care one whit about that, since to them, the only people that matter are the S&M guys (sales/marketing), so more gets offshored.
Practically every mistake in IT is recoverable, except for failing to manage customer expectations.
In particular the two ways in which I can specifically think of that lack of customer expectation management becomes a project killer are lack of solid requirements (e.g., constantly changing requirements), and mismatch between the developer's idea and customer's idea of what the deliverable should look like.
I think that the requirements one is the worse of the two because it is so easy to have this conversation:
cust: Can you just add in this one little change here? ....
dev: Sure thing
cust: While you're at it
Code Complete covers this pretty well with the analogy of building a house. "Moving" a wall is really easy when the house is just a drawing on paper. It is considerably more difficult once the foundation is poured, the walls are up and the roof is on. People building houses know that asking to move a wall in the later stages means lots of money and time on the project. However, because software is an intangible and you can't see it taking shape in the same way as a house it is much more difficult (for someone who is not a software developer) to appreciate that things that seem simple might actually be major architectural tasks for the project.
Always make sure your sales team includes a dedicated engineer. They will help keep them in line, and mitigate situations in which a customer was promised something like running Internet Explorer on AIX.
If you post as Anonymous Coward, don't expect a reply.
Not implementing and operating a data backup system properly.
I have been bit by this myself and I thought I was doing a good job at it. (I'm not an IT manager -- I'm a software engineer who often gets shoved the IT manager's job for one reason or another.)
Almost every other failure can be mitigated but not this one.
Show me an OS that can replace W10 with everyday apps. There isn't anything out there that can handle Acrobat, MS office, Outlook, Exchange and other items, other than Windows. Oh, AD capability is a must out of the box.
Don't tell me what to use. Tell me what you are trying to accomplish, and I will find a solution. That's my job as an engineer. Of course you can't run Exchange on Linux without virtualization. But if your goal is to get e-mail, calendaring, and contacts I can do that for you.
Need Acrobat? Sorry. Try coming to me and saying "I need to read PDFs" or "I need to save this document as a PDF". Sure. I can solve those problems for you.
"I want Office on Linux". No, you don't. You want some of the *features* Office provides...on Linux.
Arriving at the new hospital data center (a room 500 ft from the old data center) before anyone else (including IBM), doors open, snowing a blizzard out at 25 degrees, calling the CIO and asking him if he'd been in the room lately.
Nope.
Asking him if he had the electrical contractor's number Call it. There is not s single power outlet visible. Not one.
Yeah, they closed up the walls and painted, the electrical sub never got called.
Extension cords. Frantic 220 installs. Mangled sheetrock. The AS/400 came up about 5:30pm. I was secretly pleased our NetWare cluster was in failover...
We got done about 10:30pm on a Sunday night. No one every asked if this was an IT blunder or or a contractor blunder, but I never discussed it with the CIO , ever. He paid the overtime. My boss was litersally, genuinely speechless, a first for him. This was the same client who had a Token_Ring network that would beacon furiously on a regular basis. IBM took three months to say they couldn't do anything with the CAUs/LAMs, and they should come out and be replaced with switches. Took me asn entire afternoon to find the loose DB-9 interconnect on an 8230 chassis, the ones that were welded on back then. Bolted it in place, problem solved, we did put in the T-C switches during the move. I credit Laura Chappell, her presentation on Token networks at Networks Expo when she was with Novell, and Lanalyzer, for making me a lot of money. Thanks you, Laura!
Now there was the client who, after much analysis, believed his app vendor and replaced our 16MB Token-Ring network with 100Base-T, since they were adamant that Ethernet would outperform Token. This required recabling, drops from the ceiling, because we had reused the existing Cat 3 PDS in floor trays, but that wouldn't do for 100Base-T. No, it made no difference. The vendor them blamed NetWare and AdvantageDB, and in came the NT 4 server. The IT supervisor was the owner's son, but that's not why I questioned his competence.
I don't know how that came out because they wouldn't use us for that, we were a 'NetWare shop', despite my finishing my MCSE. Fine. I know the new guys presented migrating NetWare to NT at our Novell user group two months later. That's how it was back then. Feh.
deleting the extra space after periods so i can stay relevant, yeah.
Know what results are needed.
Does a medical database have to track phone calls, public and private databases to ensure every person who had a test got their results?
If a person had contact with a professional over a result? That person never got a result and never saw the professional after a test was done.
That a person actually got their results and did not move to another part of the country?
Bring in an expert who has worked with the exact problem around the world and who can make a database work in your country.
Real skills and the local experts get network and database they want.
Have political and gov move in and demand they be allowed to build the network with gov staff and other contractors who have no skills.
Thats how big gov can fail.
Take a project from the gov that has the skills and give to the politically connected private sector.
The contractors have no skills.
Thats how contractors can fail.
Stop using people with no skills. Stop allowing people to work on complex projects who did not pass their exams and got given a decade of social advancement.
IT can work if the right experts in the private sector, gov, mil or as contractors are found.
Stop advancing very average people with no skills into the IT sector every generation.
Find professionals that can understand complex problems.
Domestic spying is now "Benign Information Gathering"
Time and again, it's the sunk cost fallacy. A system that an organization might have spent a few million dollars to build is just not shaping up into anything they can use, but they keep at it rather than ditching it and seeing what they can do to change things.
What ironic about this is that I think agile actually encourages the sunk cost fallacy because teams will go "oohhh we can 'pivot' a little each sprint." Uh, no. If it's deep-fried dog shit for an architecture, and design you're not going to "pivot" out of this. It gets even worse when you have a management culture that doesn't understand refactoring; most of the agile teams I've been on have had managers who flat out don't care about technical debt and think they can default on it which reinforces the problems with the sunk cost fallacy down the road.
Right now this post makes me think of Ned Stark and his paper shield, except that you don't really have a paper shield, merely a piece of paper with no writing on it.
I hope to god you have some actual proof and can take it to the FTC. If you don't and you file a complaint, you might well be facing a lawsuit on any number of grounds. To be completely honest, I don't expect this to play out well for you
Don't hire a music major to be your CSO.
The best CIO I've met was a music major, he even played in a symphonic orchestra for a while. He got started in IT because he wanted to design his own DSP.
Some people are a lot more than their diploma.
lucm, indeed.
Executives with no IT experience running IT departments.
NO. More likely the executive making promises to customers and PMs without checking with the IT department and then demanding a months worth of work be done in 72 hours to one week and fire them if they can't get it done etc.
If you do get it done then it will continue. You are screwed either way
http://saveie6.com/
What you're asking is extremely unfair, especially from users who are not used to think like IT-lawyers. To them "Acrobat" is not just "reading PDFs", it is a whole set of capabilities including things you won't find in free alternatives (like encryption, form support, etc.). Asking users to articulate those needs individually, when they can summarize with the phrase "I need Acrobat", which they already know and of which they already know it supports their needs, is both unfair and unrealistic.
And it's not just capabilities, it's also a general lack of hassle. I use OpenOffice a lot, but frequently it fucks up the layout of Word documents. As a user I neither want nor need that hassle.
Look at it from a users' perspective: what they have now works. What you are selling is a journey into a world of pain, hassle, and "sorry we can't do that". Why would they want that? What tangible (i.e. not abstract) advantage do you offer to make all that worthwhile?
As soon as you outsource you are in the hands of the company you outsourced to and that company don't understand your business model, only profit for themselves.
Putting things in "the cloud" is probably the most dangerous thing you can do these days. It's like peeing in your pants, it feels warm for a little while.
If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.