The perfect IT Education? (IT/Nerd Alert)

[BushCountry]

Below is a typical Associate Level Degree program in Information Systems.   Quite frankly, I think programs like these programs are a disservice to communities they serve.  I also firmly believe that improper education and training of America's IT personal has hurt the economy and has made a major contribution to the decline in IT jobs.  Yes, I know the bubble burst, but companies are not pushing forward aggressively when it comes to IT.  So many companies got burnt by poorly trained IT personal and their empty promises that they are running scared. I value and respect the input of the people on this site, so I ask, "If you could develop/design a perfect Information System AA degree program what would it contain?" 

My thoughts are presented in italics.  Please feel free to rename courses, add/subtract ideas, and give me your two cents worth.

Introduction to computers and computing.  This course should be designed with a few thoughts in mind;  How to keep my computers/network secure.  How to maintain/optimize the units where the operator is the bottleneck.  Company-wide hardware purchasing decisions and cost benefit analysis of upgrades.  Basically, I am saying that these programs should skip 90% of what they teach about the internal operation of the computer (no text book is up-to-date and the information for the most part is laughable).

Computer programming and problem solving.  This course should be based on html, html help, java script, and SQL.

Basic spreadsheet design and development. Every spreadsheet should be designed and developed toward decisions that an IT personal makes daily.  For instance how to calculate company-wide hardware/software purchases, IT labor costs and benefits, and cost benefit analysis of upgrades.

An introduction to graphic design software.  Is this really necessary?  If necessary, this course should use software that produces flowcharts, network cabling diagrams, and how to optimize graphics for the web/databases.

Operating systems concepts; database concepts and applications.  Every computer should be a multi-boot operating system machine with connectivity issues discussed. The connectivity issues;  Security, Internet, LAN, WAN, and Terminal Services.  Database concepts and applications should relate to the current technologies for data warehousing, access times and bandwidth requirements, and backup procedures.

Database programming; installation and maintenance of computer hardware.  The database design projects should include a fictional company employee database (should allow the employee to change information as required, e.g. address info, health insurance, and dependents), computer / software / network inventory, knowledge base of common networking troubleshooting and connectivity issues relating to this fictional company, and company policy/handbook. 

Computer training and support techniques.  Cost benefit analysts can not be stressed enough.  Network and computer security, privacy and computer use policy issues, and remote administration/repair of PCs.

Systems analysis and design.  Internal structured cabling, network communications technologies, supporting remote users, firewalls, routers, gateways, and designing a secure system. 

Design and implementation of a systems project.  One design project of a new 500 workforce fictional company.  The layout of the three building complex, server software scheme and department level breakdown should be completed by the instructor.  Students should be required to make the purchasing decisions for the purchase of Network Servers, Switches/Routers, Structure Cabling and Racks,  Personal Computers, and networking/pc software.

Electives (Degree - 2 courses) - Students are required to take a 3-credit-hour humanities/fine arts course and a 3-credit-hour social/behavioral science course.  For once, I am at a loss for words.  These courses are suppose to make the IT professional a well-rounded individual.  I would like to find a more practical use for these 6 credit hours. Any suggestions?

English (Degree - 2 courses) - These courses emphasize the writing process and professional communication skills.  First course should be technical writing, practical proposals and grants.  Second course should be creating effective web documents, e.g. HTML Help and Employee Computer Use Handbooks.

Math (Degree - 1 course) - Survey of topics including sets, logic, probability, statistics, matrices, mathematical systems, geometry, topology, mathematics of finance, and modeling.  Math course topics should relate to Total Cost of Ownership (TCO) and cost benefit analysts of everything from increase network bandwidth to speeding up PC by adding memory.  In other words, practical math for the network and system administrators.

[BushCountry]

Covers IPv4 and IPv6 protocols, addressing and subnetting; Overview on switching and routing technologies; basic firewall principles and NAT translation. Network topologies and the principles of the data network stack (OSI and IP network reference models)

I can see the point of addressing the protocols, addressing and subnetting, and switching/routing technologies. Basic firewall principles and NAT Translation is a must, but how important is the OSI model to troubleshooting a network problem? I know this sounds like a stupid question, but I worked at several small to medium size companies and never once did the OSI model ever come up when it came time to troubleshoot the network (except maybe in the abstract).

[Centurion2000]

how important is the OSI model to troubleshooting a network problem

It's extremely important for covering the network issues.

Rules of thumb. If you can ping it you just knocked out 3 levels of the OSI layer. Now you don't have to check the wiring, data link layer (use switches to get the arp/mac data).

If you can bring up an FTP or web browser on a target server then you just eliminated all 7 layers and what's left is a SYSTEMS problem, not network.

I use it every day at my job. If you would like a more detailed explanation freepmail and we'll talk.

[brianl703]

TCP/IP predates the OSI layer by about a decade, and it seems to me that what you have described is basically a systematic approach to troubleshooting (which also predates the OSI layer).

Maybe one could make the case that the OSI layer is also systematic, and perhaps to someone who has no concept of what depends on what to function, it can be useful as a teaching tool.

[samuel_adams_us]

Need some electives, way more math, some C,C++, Java and .Net development classes with labs, and about 2 more years added on.

[dfwgator]

LOL. As the old consultant's credo goes, "when you see the light at the end of the tunnel, expand the tunnel."

[BushCountry]

Bump for more responses.

[krunkygirl]

bump to share with my son

[Revolting cat!]

The best IT people I have worked with had higher degrees in: Philosophy, Psychology (Ph.D), Accounting. The worst one had a degree in "Information Technology." Oh yes, he knew his IT soup du jour, whatever it was at the time, html or Pascal, or SQL. One CEO of a big corporation said in an interview I read a few years ago in Forbes that the best education one could get was a degree in Latin. Expertise in some trendy computer language, which uses like the rest of them the three or four fundamental algorithms, is trivial knowledge, in my opinion.

[BushCountry]

Something along your line of thought has showed up in several threads in this post. In a way, it justifies my belief that we are currently training our I/T people incorrectly. That is why I made the inquiry in the first place, the brainpower on this site second to none.

[Centurion2000]

TCP/IP predates the OSI layer by about a decade, and it seems to me that what you have described is basically a systematic approach to troubleshooting (which also predates the OSI layer).

True, and there was a DoD TCP/IP layer of networking model. BUT, the OSI was generated in order to teach the concenpts and to cover multiple protocols.

The other advantage of the OSI is at troubleshooting the lower levels (frayed cable, bad switch port). It is systematic and it does work well. For a beginner to truly understand networking, it makes an invaluable tool.

[Myrddin]

Some of the best CS people that I've encountered over the last 20 years were not people with a CS degree. Mike Karels was chief principal programmer in charge of the BSD 4.3 release. His degree is in microbiology from Notre Dame. I started life as a Molecular Biologist (with an FCC First Class Radiotelephone license). Keith Muller earned a PhD in CS at UCSD, but started life as a EE at Western Electric. Van Jacobson profiled and improved the TCP/IP stack by putting a logic analyzer on the CPU and comparing the most frequently accessed address ranges to a dump of the load addresses. Clearly that was a EE approach to instrumenting a kernel. I have a colleague with a PhD in Mathematics from Yale. He does some of the most interesting DSP techniques you can imagine. When he has captured the essence of the algorithm, he turns it over to me to make it run blazing fast in minimal memory.

The people who are successful are intelligent, creative, and good troubleshooters. Attention to detail, tenacity and good teamwork is important as well. Some of these traits are not "trainable".

[KayEyeDoubleDee]

Not to sound like a broken record, but are you joking? The most technically advanced field troubleshooting I've ever heard of consists of removing all the PCI cards, flashing each to the latest bios, and putting them back in one by one [with the latest drivers] until the culprit is found [and if a culprit isn't found, then you swap out the mobo]. Same principles apply to VCRs, microwave ovens, televisions, 32 bit Compaq servers, and 64 bit IBM mainframes.

I haven't heard of someone soldering a piece of electronics since, like, maybe the Korean War [unless you count all those stereophile gearheads and their Soviet era vacuum tubes].

I used to work for a company that had a third of its staff dedicated to both testing/fixing defects on new boards and modifying existing boards to suit customer needs. Typically this is done by modifying circuit-board connections between, or replacing entirely, much smaller components than you have described above. They also bring on very short term contracting to do nothing but soldering when necessary.

Admittedly those latter people [soldering contractors] are doing what amounts to purely manual labor, since they are merely told to solder point A to point B on so many boards, and not that this will allow the DSP to use this range of I/O channels, or the T1/E1 modules to support that feature, etc.

[BushCountry]

As an old micro-miniature repair technician, I miss the good old days of soldering. Repairing multilayer boards inside a hole the size of a pencil eraser was not only challenging but fun.

[BushCountry]

Bump for the heck of it.

[brianl703]

Ok, laugh if you want. Yes, I know I could buy a new one for $20, but the experience gained from fixing it is priceless,IMHO.

My old 8-port 10baseT hub which has a utilization indicator bargraph started crapping out. It was/is a problem in the power supply (bad filter caps and possibly a bad bridge rectifier which may have ruined the caps), and I will describe how I troubleshot this problem.

I first noticed it when FTP'ing files--when I put a lot of traffic through it, random LEDs in the bargraph indicator would go on and off, the link LEDs all turned off, and the blinking light which indicates that the CPU is functioning stopped blinking.

The hub has a light sequence it goes through when power is first applied. It stopped doing that shortly after I noticed that it wasn't working right.

Upon further investigation, I determined that the voltage out of the 5-volt regulator was low--about 4.5 volts, should be much closer to 5.

I tried powering it from a 12VDC power supply and it worked fine, doing the light sequence. (Normally it uses an 8VAC power supply). This confirmed that the problem was in the power supply section.

I checked the bridge rectifier, and although it was getting 8.5 volts in, only 6 were coming out. I thought that was an indication of a bridge rectifier failure so I desoldered it and replaced it with a new one ($1.39 from Radio Shack).

With the new bridge rectifier installed, it still didn't work. Further investigation was needed and was done.

I had initially thought that the filter capacitors (two 1000uF 16V electrolytics) had gone bad.

The normal way to check this is to stick a scope on the DC line and check for 60Hz "ripple". I have a scope, but I really didn't want to dig it out of the basement so I had another idea.

I thought I'd connect a normal telephone to the DC supply and listen for 60Hz hum in the handset.

This worked great. With the 8VAC power supply connected, there was a LOT of hum on the input to the 5VDC regulator. There was slightly less hum on the output. I was pretty sure this confirmed that the filter caps were bad, but I wanted to do a bit more testing.

I connected the 12VDC power supply to the hub, and did the same test. A little hum on the input to the regulator. No hum on the output. It started to look like the filter caps were really bad.

One more test: I decided to go dig through my box of parts and find a capacitor. The largest I found was 470uF, but I thought this would work to confirm the problem.

I connected the 470uF capacitor between the input to the voltage regulator and ground. When I powered up the hub, it did the normal light sequence.

From what my testing indicates, it should work fine if I replace both 1000uF filter capacitors in the power supply section.

Unfortunately, Radio Shack only sells 35V 1000uF capacitors which are too big to fit into the space that the 16V 1000uF capacitors take, so I'll have to stop by a real parts store and get them.

By the way: I have no formal electronics training (other than a course in high school, which taught me nothing I didn't already know..it was an easy "A" though). It is and has been an interest of mine for quite a few years.

[SlickWillard]

  I used to work for a company that had a third of its staff dedicated to both testing/fixing defects on new boards and modifying existing boards to suit customer needs.

I could be a jerk, and point out that that company largely went belly-up, but that would be, well, jerkish.

Obviously the free market will determine whether it's worth a company's time to repair by soldering or to repair by replacing, but in this day and age, it's real hard to see how the former business model makes sense for most companies under most circumstances.

[SlickWillard]

If you were to bill your services at industry standard rates [probably in excess of $125/hr for electronics troubleshooting (to include micro-welding)], what you've described is easily several thousand dollars worth of billable services.

An 8-port 10/100 switch can be had at buy.com for as little as $15.98:

http://www.buy.com/retail/product.asp?sku=10333167

True, you don't get the intellectual satisfaction that you get from fixing it yourself, but in the real world, a company would be insane to do anything other than replace the part in question.

[JoeSchem]

Regarding math, I read something today where a teenage girl asked, "When am I ever going to square binomials?" She's an actress, but I'm an electrical engineer with fifteen years in the business, and my moment never came.

The point is, there's a lot of math out there that gets overemphasized. Too much emphasis on the mechanics, not enough on concepts. Though some concepts like 'fibonacci numbers' are overhyped. I like the idea of more emphasis on set theory and logic.

[brianl703]

Of course; nobody would pay to fix a $20 hub. It's just not worth it. (Although I suspect it was closer to a $200-$300 hub when new; it is about 10 years old. I really wonder if the $20 switch/hub will last 10 years??)

I didn't actually spend much time troubleshooting it, I think I spent perhaps 2 hours at most, that's including the time to desolder/replace/solder the bridge rectifier.

At the least, my repairs to this hub will keep $20 from going to the ChiComs.

[brianl703]

If a replacement is no longer available, soldering is the only option.

An example would be engine control modules [ECM] for older cars. Autozone charges $100 for a replacement ECM for a certain mid-80s car. The core charge is $75, refunded when you return the old ECM.

The core charge tells me that someone is repairing these modules. Given that they are for older vehicles, they probably aren't made anymore.