back to article We trust computers to fly jets... why not trust them with our petabytes?

Listening to The Register's latest Speaking In Tech podcast got me thinking a bit more about the craze of software-defined networking, storage and whatever next. I wondered if it is a real thing as opposed to a load of hype. For the time being I’ve decided to treat software-defined stuff as a real thing, or at least as something …

COMMENTS

This topic is closed for new posts.
Anonymous Coward

We trust computers to fly jets... why not trust them with our petabytes?

The answer is obvious when you deal with the support staff of any major company (like Oracle in our case, but you could insert any major supplier here). Lots of problems and bugger-all done by "support" to actually fix them!

If that sort of company wrote avionics software (and were not regulated) jets would likely never get past the end of the runway without an incident! Nothing wrong with software in theory, just in practice its out-sourced to muppets to do for cheapness and speed (yes, right!) because they see the chance to fix it later at the customers' inconvenience, something you don't really get with hardware implementations.

5
2
Anonymous Coward

Re: We trust computers to fly jets... why not trust them with our petabytes?

Actually, it's an unfair comparison - most software which your "any major supplier" writes, has to run on a vast range of hardware, from a range of different manufacturers, on a range of different operating systems, and a range of different versions within those, interface with a range of different other pieces of software, APIs, modules, and a range of different versions within those, etc etc. Also, it's immensely complex software, due to the massive range of functionality it has to provide out of the box, due to customer demands.

Avionics software (which you compare it to) is pretty much coded specifically for, and tested on, the exact hardware (and often exact hardware revision) it's destined for, and development is significantly slower than conventional software due to laws mandating how this should take place. Additionally, due to the safety nature of it's purpose, compared to enterprise software, it's functionality is immensely simple and deliberately limited. KISS rules here.

This is why Apple have been so successful in being known for reliable software - they control the hardware as well as the whole software environment. If you can do that, writing highly reliable software is easy.

FWIW, I coded in a previous life, for a company designing aviation instrumentation, and currently work for a conventional software company.

3
0
Anonymous Coward

Re: We trust computers to fly jets... why not trust them with our petabytes?

because a plane and it's SOB isn't as important as an Olympic size swimming pool full of data (or summink)

0
0

Indeed..

<< Nothing wrong with software in theory, just in practice its out-sourced to muppets to do for cheapness and speed (yes, right!)>>

You beat me to it.

I really didn't realise just how bad it can be until I started on my latest contract; I have had to deal with out sourced programming work before and was fairly non-plussed by the quality and speed of their work, but these people take it to a whole new level below the basement car park.

When you look at it, the company really have not saved anything; the inefficiencies of the current system are costing them serious amounts of money. Getting the right people in house, even at a realistic wage for the UK would cost a bit more, but the work would be done properly and within the right time frame and this would save a great deal of time and effort (i.e. money!) across the whole business.

But it won't happen - why? Because the senior management only look at the single figure of cost and despite the fact that they are supposed to be taking a strategic view are so myopic that they make Mr. Magoo look like Hawkeye.

5
0
Gold badge
Unhappy

Re: Indeed..

"But it won't happen - why? Because the senior management only look at the single figure of cost and despite the fact that they are supposed to be taking a strategic view are so myopic that they make Mr. Magoo look like Hawkeye."

Sounds likely.

I'd love to add up the cost of every change order and bug fix a piece of outsourced software costs over it's lifetime.

With my software-houses-are-machines-for-producing-sofware-faults hat on I'll guess eyewatering relative to the original project cost.

1
0
Silver badge
Thumb Down

Re: Indeed..

"But it won't happen - why? Because the senior management only look at the single figure of cost "

Actually, there's another reason, and that is information asymmetry between the two sides of the make or buy debate. Typically the marketing budget is around 4% of turnover for a large company. In the UK tech sector, outsourcing is worth around £50bn a year, and that means that the vendors spend around £2bn a year on marketing and promoting outsourced solutions. Then there's the useless, useless parasites of the management consulting firms (a market around £7bn a year), who promote outsourcing on the basis that they hope to pick up implementation or advisory contracts, and often dress up their marketing as "research". All told there's probably around £2.5bn spent in the UK pushing what a wonderful, risk free opportunity outsourcing and offshoring are. Now what's the marketing budget of the in house tech function, and how much clout will it have in the boardroom?

So the directors will be on the receiving end of a never ending stream of smooth talking salesmen, pushing the idea that the in house function is slow, expensive, ineffective, and that their product is fast, cheap, and generally brilliant. The CIO can try and fight that flood of nonsense, but what's the chances he'll either try, or succeed, when the CFO believes the guff Arsenture publish, or the CEO has been captured by the Bollox Consulting Group? These people turn up with powerpoint slides showing how much cheaper each employee is in some sweaty crack of the planet, and with "research" in which the CEO's peers (supposedly) all admit that their priority is offshoring to capture wage arbitrage. So the drip-drip-drip is "Everybody else is doing this, and you'll be left behind if you don't, we can help you avoid looking stupid. But if you don't do as we say, everybody will laugh at you".

In IT we refer to this as a social engineering attack.

0
0
FAIL

Not convinced a market can handle this requirement yet

Markets are wonderful resource-allocation mechanisms in environments where we have backups due to their chaotic nature. But this article proposes we entrust our backup storage to a market, where vendors with cash flow problems are likely to tell untruths, especially when questions are asked about their reliability, and we want computers to be able to see through such lies made to them by other computers ?

MegaUpload showed us that cloud storage is not inherently as reliable as storage we manage in house. What this proposes seems much worse, but no doubt we'll be seeing plenty of purveyors of emperor's clothing to cost reduction specialists who will be in another job in six months' time where they will be unaccountable for their current decisions.

0
0
Facepalm

We trust computers to fly jets

What?!

Do we?

1
1
Happy

Re: We trust computers to fly jets

The computer's easier to drag out of the airport bar and might not have overdosed on prescription amphetamines.

0
0
Silver badge
Boffin

"We trust computers to fly jets..."

But we do not trust them to fly without human back-up, do we?

And even then, if the humans misunderstand what the computer is telling them or get the settings wrong, we get disasters like the Air France crash off Brazil.

5
0
Anonymous Coward

Regulation

We trust computers to fly jets... why not trust them with our petabytes?

Because if you compare the type of regulation in aviation with that in computing, you will find that they are diametrically opposed. And even aviation gets it wrong sometimes.

4
0
Silver badge
Linux

Re: Regulation

Heathrow may be an auto-pilot only landing zone but there are still two real pilots sitting behind the yokes just in case something goes sideways.

0
0
Silver badge

I seem to remember reading in PC-Zone 'back in the day', about a flight sim enthusiast who got to look at the cockpit of an RAF jet for real. He gleefully noted that he knew what all the knobs and switches did, bar one little red micro-switch... "Whats's that?" he asked.

"That's to reset our computer when it crashes" the RAF pilot told him.

2
0

Three finger salute...

there's even a three fingered salute for some mission computers... a combination of three hardware switch inputs that will never normally occur together to be used in the event of it freezing. Monitored on the hardware level in simple logic circuits and will force a cold restart when they occur.

0
0
Megaphone

"We trust computers to fly jets... why not trust them with our petabytes?"

Cost and safety/mission criticality.

90% of the code used in aircraft is there to detect when something goes wrong, this code is also implemented in different ways across different platforms with the results analysed and double checked.

There are very strict standards that must be adhered to for the code to be even allowed to go anywhere near a plane and then it needs to be further tested before it is allowed to fly.

As stated above, this is not infallible and still subject to human error.

The cost and time to implement these systems is astronomical compared to a PC operating system based application.

Is it possible to do it? Yes. Will a business do it, probably not, simply as the cost and timescales would put them so behind the market that the business would be bankrupt before they sold license 1.

4
0
Bronze badge
Thumb Down

What kind of a comparison is that?

I suggest you go away and read DO-178B. Thoroughly.

It's a start.

1
0

Do we?

As far as I recollect, we do not trust computers to fly jets. There are usually three computers and the output of each is compared with the other two. If one of them is different, then a human is called in to fix the problem

0
0

It's called a database

When will DBMS take in as input both SMART level data from the drives, as well as policy level decisions about the potential cost of losing specific datum?

0
0
Anonymous Coward

Continuous revolution, not evolution

Commercial / consumer IT is driven by the need to generate revenues - for product vendors, by continually churning their base with new versions of their products, complete with new bugs; and for service providers, by continually "transforming" the services they provide. Continuous revolution, not evolution.

So, unlike aircraft which a) are developed with rigorous engineering processes; and b) which have a lifetime of decades, during which they undergo continual improvement, consumer and commercial IT products have life-cycles measured in months and the race to get some new glitzy feature to market overrides all other concerns.

The sum total of i) the costs of continually developing product changes, ii) the costs of the IT service industry rolling those changes out, and iii) the costs to businesses of having to absorb the continual change inflicted on them by vendors; must be astronomical, enough to develop products with robust engineering methods that go some way towards providing the reliability and security that is increasingly necessary as we become more dependent on these technologies - even when on the ground.

2
0
Pint

Cant see that working at home but

I bet Amazon S3 does some of these things already

0
0
Anonymous Coward

Nearer the real root of all storage evils

Software/Firmware will either be reliable or not dependent on the quality of development and specifically test cycles in that development exercise, nothing new there.

Interestingly there is a leaning towards some form of data classification with this approach, at least in terms of performance and value.

However it does nothing to ensure quality, which includes management of data instances - version control for instance because I've spent hours confirming which version of a document is valid, and regulatory compliance - like it or not more is coming and the *real* challenge will be to define what policies need to put in place to meet them.

The good news is a software stack to manage on the basis of performance and value is a potential base for implementing regulation driven data management.

0
0

Actually there are some orchestration engines that do allow storage to be added as a set of parameters and then deployed against application needs.... HP Orchestration Operations is one of them. It enables the automatic deployment of servers/storage for applications and VMs

0
0

Seen the announcement from Nexenta on using the lib storage management API? They join NetApp, Red Hat and a few others in developing an extended open standard for storage management and discovery. Extending to your vision seems logical - and only slightly beyond the stated project scope.

Most data placement decisions are based on simple rules such as optimize IOPs, optimize $/GB, group together, snapshot schedule, and geographically replicate.

0
0
This topic is closed for new posts.

Forums