Red Hat exec proposes end to IT suckage Why isn't IT getting cheaper instead of more expensive? That was the question posed by Jim Whitehurst, president and CEO of open source software juggernaut and cloud wannabe Red Hat, during his keynote at the Interop networking — and now cloud computing — trade show in New York. After all, processor and storage capacity keep …
Well, what do you expect in an industry where the two biggest competing architetectural approaches do not take performance into account during the standard requirement gathering excercises.
If a structural engineer was designing the Dartford Crossing bridge using UML it would have been designed for one car because the requirements would have stated that the actor requests transportation of a vehicle from one coast to another. Actor. One. Not 50000 of them in peak hour.
If a structural engineer was designing the Dartford Crossing bridge using Agile the bridge would have been launched to customers without the tarmac first. The tarmac would have been added on first iteration after customers complained. The banisters would have been added on second iteration after a few cars go off the bridge. And so on.
In both cases the bridge would have been able to pass only a car at a time or a few after a number of hotfixes by "red" teams because none of these processes captures _END_ result performance requirements as a part of the standard process (some companies have special extra processes for that, but that is not part of the normal agile/UM as evangelised by their respective inventors).
Not that the other "newer" architectural frameworks are any better either.
...you do know that designs are modelled and tested in a computer before use, don't you? Scale models (or prototypes) made and tested (often to destruction). These cycles repeat umpteen times. THEN the final item gets made.
Also, there's a few thousand years of empirical evidence to draw from.
And with all that....it can still go wrong. Sometimes badly.
Show me how many software companies do a proof of concept for a system before committing to doing it properly? - Very few
Show me how many software companies do a proof of concept that is used for a _PERFORMANCE_ and _SCALABILITY_ verification? In an actual test harness? Well, I have yet to see one.
So it is indeed one car going across the UM bridge because that is what the actor/action/process interaction describes and a few cars going off the agile bridge because there are no railings yet in this iteration.
Anton Ivanov:
"Show me how many software companies do a proof of concept for a system before committing to doing it properly? - Very few"
Plenty, actually. Some agile approaches call these "spikes"; I've participated in a number of them.
"Show me how many software companies do a proof of concept that is used for a _PERFORMANCE_ and _SCALABILITY_ verification? In an actual test harness? Well, I have yet to see one."
Well, if you haven't seen one, they must not exist. Which is odd, because I'm quite sure we do performance and scalability testing, on prototypes and product lines, using dedicated test harnesses. Hell, I'm pretty sure I helped build some of those harnesses and develop some of those tests.
"So it is indeed one car going across the UM bridge because that is what the actor/action/process interaction describes and a few cars going off the agile bridge because there are no railings yet in this iteration."
It'd be a lot shorter, and just as correct, for you to write "I have no idea what I'm talking about, but I think the UM and some unspecified agile process produce ridiculous results".
"We are not selling them technology, but partnering to jointly build the solution. We're not building it — they are"
followed immediately by
"The key to unlocking the value of clouds is open standards for cloud interoperability, says Whitehurst, as well as standardization up and down the stack"
Yes because customers will happily change the way they want to do things in order to be more similar to the next guy. Anything to reduce work for the poor coders who are always top of mind.
Of course you don't need smart people to build the systems. Turnips manning the keyboards directed by legions of managers invariably produces nothing but the best of sausage.
Etc.
"Whitehurst didn't mention that this waste provides $500bn worth of jobs, and that these also have a multiplicative effect, so it could be a wash. But I digress..."
That's the broken window fallacy. It ultimately never contributes to economic efficiency, though economic systems can be sufficiently complex that paying people to do nothing useful can appear beneficial in the *short* term.
lambasting Novell for their "expensive to manage" Netware solutions. Yes it took highly skilled staff to run an infrastructure, but in 1996/1997 we were running servers with load volumes (users and print, not volume of data of course) that we didn't reach on Windows until 2003... and the comparison is similar to Directory (AD *might* have finally caught/surpassed 90's era NDS with 2003).
When we started migrating everything to NT I'd say conservatively we doubled our staff and might have tripled our server counts. You couldn't even assign quotas to shares and the Domain structure of NT was definitely not geared for Enterprise.
To say nothing of hardware, software and utility costs... I never understood why anyone saw this move as an improvement unless Novell was severely overcharging.
At the end of the day, IT shops tend to be more of a boat anchor than an enabler.. but our partners on the business side share in the blame as it's most often a dysfunctional relationship between the two, not necessarilly a dysfunctional IT shop, that's the real cause of the problem.
"Whitehurst didn't mention that this waste provides $500bn worth of jobs, and that these also have a multiplicative effect, so it could be a wash."
That is the IT version of the "Broken window fallacy" and it would have done his credibility no favours whatsoever had he have mentioned it.
"The problem, says Whitehurst, is that IT vendors try to predict the future many years out, and design their products for big-bang upgrades far into the future"
He is dead right there. I have been saying much the same thing for years. IT vendors have yet to step out of their glory days (aka the 80s) where every site was a greenfield site and every customer a locked-in income stream.
The needs of the market have matured since then but the entrenched players haven't kept pace.
future proofing of their software and infrastructure so that when the next big thing comes along (for example delivery of TV over Internet) they won't need to go out and buy a set of brand new software and hardware.
Whilst I don't always agree with the way software and hardware is sold to customers by the big companies, the reality is that much of the functionality in their products is there because its what customers ask for. Pretty much all the software vendors have customer forums which provide input in to their development lines so that they can create new functionality to meet the customer's requirements. Reading this article it appears to say that no new functionality is required until the point in time that the customer wants it and that open source can deliver that functionality at that point in time at minimal cost. And that is a claim I cannot believe.
The answer is simple, people drop old (perfectly serviceable) hardware. Win7 (one example) dropped support for lots of things. OEMs like this as they can sell more kit, so they rush to bundle Win7. MS likes this as they can ship more to the great unwashed, so they deliberately drop support to force people to upgrade. A virtuous cycle for consumption.
This is true for end-users and for corporates as well. In fact it's true of many things in life, but it shows up most in IT.
It has to stop. Not just for economic reasons, but for environmental too. We need to find ways of supporting, re-purposing and recycling older kit.
I am fed up with the likes of Nokia preaching to us that we only use 40% of our smartphones' abilities when only 40% actually works properly. Most sold software sucks because they need an ongoing revenue stream from upgrades. In order to persuade us to upgrade we get stick and carrot. "Shiny new features" and "End Of Life" messages assaulting us daily.
That does lead us to the question that if each new version is so much better than the last, just how bad was the first version and should they have been allowed to sell it in the first place?
There are effectively no consumer standards for software. If I buy some hardware, it has to comply with the low Voltage Directive, the EMC directive, and so on. However, I can buy a Nokia phone that claims IMAP4 compatibility, and find that it only supports half the commands in the standard. At best I can get my money back, but I cannot enforce full SMTP or IMAP or HTML4 compatibility. Doesn't matter whether this is cloud-based or hand-held. We all rely on software, and we are consistently sold short. Over-promised and under-delivered.
As to the cloud being the solution to everything, it isn't. Symmetric broadband speeds are only just getting to 10Mb for under 5 figures a year, yet LAN speeds are already looking at 10Gb/s, having left 10Mb/s behind a decade ago. Try telling a CAD or graphics company that the cloud is the answer to all their problems. I think not.
It's been done (site-ready for the user count you like, too) all right; the app is of course given user privileges rather than a browser-cache and a cookie's chance at life, but Service is solidly in the delivery options. You don't even need Steam running first...oh, you should, but you don't need it any more than a blade farm just for tipping your models back up. Unless you don't look phone e-mail in the teeth past winging a 1GB attachment. Then you probably need a bladder server and a fresh teeth server.
The AutoLISP console looks freaking awesome on an iPad, but you'd better have crammed some RAM in there if you want much control over an all-systems walkthrough.
Of course we'll be thrilled to see what part of Veritas Suite or whatforth Red Hat is going to crowdsource and authenticate trainmasterwise for us.
"Whitehurst didn't mention that this waste provides $500bn worth of jobs, and that these also have a multiplicative effect, so it could be a wash."
Not having waste wouldn't necessarily mean fewer jobs -- the mythical "no waste" IT department would be getting projects done on time and on budget, the projects would run smoothly, help make the work of the rest of the company just a little easier and so help the company make more money. Rather than doing what they do now at a lower budget and lower number of employees, they could instead be authorized to do more projects to make the company run even better.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
