IT, Testing & Best/Worst Practice [split]

[Ent]

Content removed by poster

[Son of a Beach]

more testing to go.

You do work in IT? Test is not a word I thought IT was familiar with

Cheers Brett

Heheh... I don't know which IT firms/departments you've been involved with in the past, but where I work, every significant system has an exact (or as near to exact as feasible) replica on which all testing is done, and only when changes on the testing system are thoroughly tested, can they be approved for implementation on the production system (well, that's the theory anyhow - we usually stick to it but there are the occasional lapses). Some systems have more than one replica - sometimes another one for development, and sometimes more for different purposes. It makes for very expensive IT infrastructure, but the value is worth it to avoid problems happening on real live production systems.

I've been involved in IT long enough to have witnessed the occasional catastrophic effects of applying changes to production systems without having put them through their paces in testing environments first. So I try to apply good practices to this site most of the time too (I don't test every little change, but anything significant spends a bit of time in the testing environment first).

[Ent]

Content removed by poster

[photohiker]

To be fair, lack of testing infrastructure is not a Microsoft or Navision issue. It's generally a lack of understanding and an unrealistic budget that delivers that kind of mess.

Even when we ran Pick, we had a test box. While it can be daunting (and expensive due to the various license issues) we ran a big name business system on MS SQL with Prod/Test/Dev

If you have 'consultants' delivering changes directly into Production they seriously need a riot act reading session, although I suspect that at this stage you need to re-educate the whole of management because it never gets to this state on it's own.

[Son of a Beach]

Yep, even with all our 3rd party applications (including some MS ones), any upgrades or other changes are all thoroughly tested before anything is touched in the live production environment.

Its really up to the IT staff to learn a bit of good practice, as in IT, everything is complicated, and any change can result in unexpected behaviours.

PS. Perhaps I should say especially for the MS ones, as they do not have a reputation for quality.

[ollster]

The thing about IT is that these days we have such high uptimes, people get really upset when there is the slightest disruption to service. The cure is obvious, we need to go back to more primitive times when the words "test", "disaster recovery" and "back out strategy" were unkown. Then maybe people will start to appreciate what they've got!

Good job Nik.

[Singe]

more testing to go.

You do work in IT? Test is not a word I thought IT was familiar with

Cheers Brett

Sure it is - we test stuff by rolling it out - the users quickly let us know if anything doesn't work

[Ent]

Content removed by poster

[ollster]

TLDR. Sorry, I tried, I really did.

[photohiker]

With Navision, leading MS accounting system, just about every standard report does not work or is so basic as to be useless, hey why should a stock report have a total and trail balance do not either as standard.

A couple of points.

1) MS bought Navision, they didn't write it.

2) Reports on ERP systems are one of the core implementation profit centres. Why would you kit out your system with production ready reports, are you crazy? You can easily budget $5k for a report. One of the systems I have been involved with (not Navision) introduced a new whiz gui report writing module that was so complex and filled with little hidden traps that it actually cost more time to create a report in the GUI than the old way (basically header/body/footer field x,y location, length, data source, modifier, formatting)

Standard line in sales presentations was to show some standard reports but not to dwell on them. (if you do, people will notice they are crap) Just include the time per report x the number of reports identified in the quote/scope.

Yes, I am pretty cynical of the process. The client is always the loser because they are between a rock and a hard place. Even if they know anything about the system (and usually they don't) they are the ones with everything on the line and everything to lose if it goes pear shaped.

these days we have such high uptimes, people get really upset when there is the slightest disruption to service.

Dunno what systems you have been looking at, but I have never seen anything but high uptimes. The reason people get upset is that unreliabilty ruins their businesses. Too many service providers regard short disruption to service as 'minor' without understanding the repercussions.

[ILUVSWTAS]

My cats breath smells like cat food

[ollster]

My cats breath smells like cat food

Sigh. At least you understand levity ILSWT... people are way too serious on this board.

[ILUVSWTAS]

My thoughts exactly brother

[corvus]

The thing about IT is that these days we have such high uptimes, people get really upset when there is the slightest disruption to service. The cure is obvious, we need to go back to more primitive times when the words "test", "disaster recovery" and "back out strategy" were unkown. Then maybe people will start to appreciate what they've got!

Good job Nik.

Boy I must be a natural for starting wars Back when I was young and kept a T-Rex as a pet and Unix, IBM and Digital ruled the world programmers and analyst were highly trained people skilled in testing regimes such as test first, middle and last occurrence as that will nail 90% or more of potential problems plus compile times were so long that you wrote code carefully and were annoyed when it gave a compile error or did not work. Also been text based operating system a technician needed to really know their way around just to do basic things so meant it was not such a stretch for the trickier feature or problem. None of this click box approach.

As for up time try this, six years on a Unix system until hardware failure brought it down. Foxbase application written in 1987-1990 for stock system that has two on site visits. One to do year 2000 compliance audit (set century to 1950 was the central configuration change) and shortly afterwards for the "never ever GST". The business despite my repeated requests refuses to replace it because nothing they have seen can match the speed, reporting and reliability. In fact when they visit test sites the test site personnel have come back and ask for contact details of their system provider. Try remembering code and programs from twenty-years ago In fact I am certain if I wind up in a dark damp cell kidnapped by an armed terrorist group they would arrive to get me out to avoid changing systems. Only trouble is about five years between phone calls so could be a long wait

And the above would be considered just acceptable performance by the big end of town programmers from Synergy. It was interesting when their programmers read the changing winds towards MS domination and started to skill up in MS they found most of the standard "C" code functions from MS that their Database relied on had bugs and could not believe that their test servers would self destruct after six months with MS explanation been you should shut down and reboot your servers periodically and not leave them on. These are people that struggled to accept that Unix was stable compared to IBM operating systems. They only every saw an IBM mainframe or System 38 shut-down when it was physically moved Software failure generally meant programmer's coffee escaping into a keyboard.

Test and data recovery strategies were standard operating procedure with often a "test" server off site running ready to have live data restored to it from tapes. And this was done weekly as a matter of course in case drive head alignment issues crept in. In fact I was horrified when one "leading" Melbourne provider of our new accounting software had no idea what the mentioned strategies meant in real life. Yes they appeared in the nice thick sales document as words but I could not see anything how they worked so pushed for answers. The give away was no test environment and a sheepish comment that "old hat man" you do not need it and sales section should update the document

The record I am personally aware for uptime was a Unix box put in 1986 (IBM model 80) that failed in 2004 when the dust finally chocked the power supply. The installers had automatic disk clean-up programs run by cron and fixed purging of history so the system played nicely within its allotted space for all those years. In fact the company "lost" the server when the space it was in was sealed off in office renovations in the 1990's. Um? tape backup was interesting as all the magnetic media had fallen off. The error message as legend had it came one day but as the sky did not fall in the standard procedure was to ignore it. Now that was class programming and analyst work by the designer who had long since retired to a beach somewhere in the late 1990's plus testimony to IBM approach in building computers. Would like to shake that person's hand even with him been a South Australian

It probably will be the unavailability of tractor form stationery that will be the undoing of the old Pick and Unix system rather than anything else. With Navision, leading MS accounting system, just about every standard report does not work or is so basic as to be useless, hey why should a stock report have a total and trail balance do not either as standard. Heck the skilled professional gold standard installer did not even have the GST setup in half the companies and when I finally got to look at the in built installation check list it was not used despite been MS "stated" installation approach. The much vaulted dimension feature has more holes than Swiss cheese. Do not mention Greentree to a few sites as they have just come out of therapy as it is possible for a "developer" to write applications that you can not actually get the information out, ever, or more accurately by a report as each record can be constructed differently. What I mentioned above is very common with MS solutions regarding set-up problems. Yes there are good ones out there but finding them is near impossible and thankfully I managed to get on board a good one in the end so much of the above is now history.

Now I must go as my computer has done an updated and wants to reboot and besides this probably even by my broad standards can not be linked back to bushwalking

So endith the rant.

Cheers Brett

PS top job Nik and glad you have a "proper" IT environment and do not work for a bunch of cowboys that haunt the mid range of applications.

Cheers Brett

EH!!
corvus

[ILUVSWTAS]

[Son of a Beach]

(This topic has been split off from the phpBB 3.0.6 INSTALLED topic.)