Why Your IT Spending Is About To Hit the Wall

CowboyRobot writes "For decades, rapid increases in storage, processor speed, and bandwidth have kept up with the enormous increases in computer usage. That could change however, as consumption finally outpaces the supply of these resources. It is instructive to review the 19th-century Economics theory known as Jevons Paradox. Common sense suggests that as efficiencies rise in the use of a resource, the consumption goes down. Jevons Paradox posits that efficiencies actually drive up usage, and we're already seeing examples of this: our computers are faster than ever and we have more bandwidth than ever, yet our machines are often slow and have trouble connecting. The more we have, the even more we use."

Comments

I know what you're talking about

[SpaceCadetTrav]

Despite technological advancements, it takes forever for Slashdot to load on my phone.

Re:I know what you're talking about

[MachDelta]

Yeah, they broke^h^h^h^h^h^h improved the comment system a while ago. In the name of progress, of course.

Re:I know what you're talking about

[Anonymous Coward]

In light of technological advanvements, better bottlenecks are being implemented.

Re:I know what you're talking about

[hairyfeet]

Which shows it isn't the OS or the hardware, its the networks. i have to wonder if the whole "buffer problem" we've been reading about here is about to hit the tipping point as it seems like everything now has big fat buffers built in and as we know the Internet model simply wasn't designed for having large buffers throw the timing all to hell.

Personally I think it is high time we use an old solution to fix a new problem...bring back the WPA. a lot of our bandwidth problems would disappear if we had nationwide FTTH or at least fiber to the neighborhood. It seems like a great way to put all those sitting at home on unemployment to work and you build it right and just as many bridges built by the WPA in rural areas still work fine so too could a well built fiber network last us for ages. this would also give us the benefits of new businesses springing up to make use of this new resource and finally kill the duopolies that have been hamstringing growth in so many areas of the country because the lines would be open to competition.

I truly believe if we don't do something radical like bring back the WPA we will end up staying on the short bus to the info superhighway because the corps can make more money by throttling and cherry picking than by actually growing their businesses and in our short sighted corp climate the quarterly reports are all that matters. i know that even though my home town has grown by more than a third neither the cable nor DSL has moved an inch in a good decade or more. They would rather just add caps and sit on the big wads of money than actually add new customers. If we don't change this situation we are gonna end up being left behind so its high time we put those unemployed to work building us a new broadband infrastructure.

Re:I know what you're talking about

[Anonymous Coward]

No it's not the networks, it's the morons in charge of the content.

Slashdot should load 10X faster than it does, but the uneducated developers and designers put in a lot of crap that is not needed to add in "pretty" that does not add to the content at all.

So slashdot now takes over 10X in bandwidth and processing power to deliver the same content it did 8 years ago. All so I can gave some web 2.0 crap that does nothing at all.

But it's not just slashdot. ALL websites are bum rushing the add more crap idea. Facebook takes 10X longer to load from 5 years ago, CNN, ESPN, etc.. all of them have went from hiring competent people that understand that adding more data to send to the viewer is bad , to a bunch of morons that use every JS toolkit known to man so I download 40mb of libraries before the page loads. Some JS is useful. Good programmers put in the libraries only what is needed, posers put in the whole damn library. This same trend is on Desktops and phones. android and IOS suffer from this as well.

It is about to hit the wall because low paid low skill developers are what companies hire compared to highly skilled people that will do it right.

Re:I know what you're talking about

[jmorris42]

> But it's not just slashdot.

No it isn't. If the average visitor isn't impacted the devels don't care. But if the average user were impacte dthey would. Which is the problem with the concept under discussion. The belief that bloat MUST be therefore there being nothing that can be done we are all doomed to spend ourselves into poverty fighting a problem that will never exist.

Because as soon as it becomes a problem, suddenly the average pageview will suddenly be able to shrink in half without impacting usability at all and if that doesn't do it it can cut in half again with minimal impact. And it isn't just webpages, most everything suffers the same bloating. Does a simple little game that was a 50K download on Palm OS really need to be a 1MB app on Android or iOS? Nope. But because users don't care the developers don't care either. And again, if the first part of that statement changes you can bet yer butt the second one will.

Short version: This is a self correcting non-problem.

Re:I know what you're talking about

[crutchy]

of course everyone loves eye candy, even at the expense of usability and stability

Re:I know what you're talking about

[crutchy]

programmers can keep trying to make their software idiot-proof, but society will forever continue making better idiots

Re:I know what you're talking about

[Jane Q. Public]

"No it's not the networks, it's the morons in charge of the content."

In many ways I must agree. I have strenuously protested many of the changes made to Slashdot over the last couple of years, which have seemed to add nothing substantial to usability, and instead have added overhead and time, and actually made it MORE difficult to use.

Re:I know what you're talking about

[dgatwood]

Does a simple little game that was a 50K download on Palm OS really need to be a 1MB app on Android or iOS?

Depends on the app. If we're talking about an all-text game, that's a little extreme. On the other hand, if it contains any image assets at all, that is probably not unreasonable.

Remember that the original Palm hardware had 240 x 160 resolution in black and white. A current-generation iPhone has 960 x 640 resolution in 24-bit color, and it is usually bundled as dual-platform for iPad, which is 2048 x 1536 in 24-bit color. So if that 50k app on Palm were nothing but uncompressed image data, you would expect the iPad/iPhone version of the app to be a whopping 96 megabytes.

Obviously image compression helps with that, and obviously an app contains content other than image assets, both of which contribute to that being something of an overestimate. That said, using that as an upper bound, a mere one megabyte doesn't sound bad at all.

Re:I know what you're talking about

[HapSlappy_2222]

While you're absolutely correct, the solution is not to re-train developers to "do it right"; it's really not feasible, and isn't a final solution by itself anyway.

Just like has been predicted for years, IPV4 is running out of addresses. We *could* force the A block owners to give up their IPs, we DID kludge in NAT, but the proper solution was to increase the available resources. For bandwidth/processing power issues, we need a multi-pronged approach, including increased resources (infrastructure upgrades), better development (cleaner standards, maybe?), and competing providers. I like hairyfeet's suggestion for this, though I think it's a bit of a stretch to present fiber to the home as a way to solve unemployment, too.

Re:I know what you're talking about

[dgatwood]

Unfortunately his plan was to turn-off free TV (all channels 25 and up) and turn it over to wireless companies. That's not a solution... at least not as good as Fiber to every home.

It's a fundamentally unworkable solution to the problem. The reason we don't have enough capacity is not because we need more bandwidth. The reason we don't have enough capacity is that we're trying to use one tower every 15-30 miles to provide service to hundreds of thousands of people. If those folks are mostly using it occasionally (as they do with cell phones), it works reasonably well. When they're sitting there for hours on end surfing the Internet at home or work, it breaks down very badly. We're orders of magnitude away from being able to handle that.

Wireless works really well at short distances where each cell is talking to dozens of people (e.g. Wi-Fi). The larger the number of people per cell, the more infeasible it becomes due to interference from other devices, not to mention all the multipath problems inherent in wireless delivery over long distances. Even if you could make the bandwidth ten thousand times wider, we still wouldn't have enough bandwidth to service every man, woman, and child's home Internet needs somewhere like New York or San Francisco using cell towers. It's entirely the wrong solution to to the problem.

Instead, we should be focusing on making VoIP and VPNs roam transparently between cellular services and Wi-Fi, roam transparently between multiple Wi-Fi hot spots, etc. And we should be moving more towards providing free public Wi-Fi services at high densities so that only the last few feet are wireless.

Re:I know what you're talking about

[hairyfeet]

Sorry I didn't elaborate but I have been accused of making posts too long so i try to shorten my responses. what i meant was not ONLY for the WPA to be running FTTH, and as i said even FTTN would be a huge improvement in MANY areas, but there is so much infrastructure that can be made a thousand times better. just look at our roads and bridges, many are from Ike's time. We have also seen there IS a way to build a road so it will really last, just look at the Autobahn, but you have to lay a really solid foundation and build up.

I think we can use a modern WPA to truly transform this country into the vision that many of us were shown as kids in the 60s, with truly modern roads that don't break easily, could have embedded sensors for future driverless cars, bridges replaced with better designs, and of course by building FTTH or FTTN depending on the size of the area to help bring tele-education and telecommute to the masses.

As a final change I would toss this electric car nonsense as the battery tech isn't there yet and instead build a true "people's car" which would be both a 2 door and 4 door model that had a 4 cyl and gets 40MPG for less than $10k. We would then offer a cash for clunkers style program so that all the working poor could trade in those old gas hogs for a much better vehicle for the environment.

with these changes I think we could lick unemployment and modernize our country for the future while at the same time lowering our dependence on gas from overseas and at the same time actually creating jobs for all those out of work. sadly though it requires vision and will, two things our politicians seem to have little to none of ATM.

Re:I know what you're talking about

[gmack]

It just isn't possible to build a road that really lasts as long as the contractors know that if the road breaks next year they will get payed to fix it. The system needs changing to one where the contractor guarantees the work for x number of years so they aren't motivated to pull crap like overheating the asphalt to save trucking costs (makes the road last for a much shorter time).

Re:I know what you're talking about

[silas_moeckel]

Fiber networks are fine and dandy but you need to make it an all optical passive network. CWDM is cheap and works well and you can always upgrade to dwdm if you need more to a single drop. I'm talking about at least a fiber to each house with the municipality or some other owner being responsible for the glass only. Anybody that wants to can lease space in the CO or fiber to the CO (I would expect a big market for CO to CO fiber would spring up). Let them deliver anything they want internet, phone cable, metro ethernet, etc. Once you have a clean divide between what needs to be a monopoly and whats need not you can get real competitiveness. Think of what could happen if I could order a lamda cross connect for no more than 4x the cost of any of those services for an in town link.

PS this would also get those fugly boxes off the telephone poles and rights of way, all fibers go all the back to the CO.

Re:I know what you're talking about

[TheCouchPotatoFamine]

is it windy on your high horse? Fact is, computers exist to be used - and if the programmer isn't using the hardward to facilitate a pleasurable experience (for the developer or the user, actually, since happy devs eq. bug free programs), they might as well be wasting the consumers time/computer. In short, cycles spent on abstraction are the best spent cycles, because that's what computers are for.

Re:I know what you're talking about

[hairyfeet]

Nice to see i'm not the only one. I mean if it is a complex subject how is one supposed to argue for or against something with the space of a tweet? Frankly i think the whole tweet thing is helping to make discussions into basically "it roxorz" or "it suxorz" because that is all you can really fit.

And I guess i'm not good at explaining what i'm suggesting as you are getting the wrong idea. I am NOT suggesting that unemployment be tied to these things but what I AM suggesting is this: You have all these people that are already unemployed. We know that unemployment breeds crime, it hurts people's self esteem and ability to get a job, and it causes real hardship...so why not put these people to work? Giving them a job they can be proud of, something they could show to their children and say "I helped to build this" would be better than just having them sitting at home with a knot in their gut wondering if the unemployment will run out before they get another job, don't you think?

I just think by setting realistic goals, unlike this electric car nonsense which as another poster wrote is like saying one could build a 747 in 1906 if you threw enough money at it, but instead to take a resource we already have that is going to waste, all these unemployed workers, and instead utilizing them to do jobs that really need to be done. I will admit you can call me a little biased since i have walked across bridges that my grandfather helped to build in the WPA but I truly think these goals are obtainable with what we have NOW.

Just as i believe a "people's car" with a price that can be reached with the technology we have and a MPG goal that can be obtained without any new technologies having to be invented, could truly change this country for the better. the last report I read said the average age of a car on the road in this country now is 11 years and the average MPG is barely 20. Now imagine if suddenly everyone is driving 40MPG cars because anyone can afford them? Not to mention I have no doubt like with the Model T we would see entire new industries spring up to customize these new vehicles which would also help our unemployment situation.

So I'm not saying tie one to the other, what I am saying is we have all these jobs that need doing, we have all these people that need work, why not put them together? I think we can all agree our infrastructure needs a major overhaul anyway and we are falling behind on broadband compared to other nations, so why not instead of paying people to sit and worry we do something about this? And when it is all over if there is still major unemployment at LEAST you will have fixed what is broken and this new infrastructure and telecommunications grid will encourage other companies and countries to give the USA a second look. And who knows how many industries that we can't even imagine would spring up from a nationwide fiber to the neighborhood system? after all look what happened when we took the Ma Bell monopoly away, suddenly we had faxes and wireless phones and this system we are typing on now!

Re:I know what you're talking about

[Tadu]

We have also seen there IS a way to build a road so it will really last, just look at the Autobahn, but you have to lay a really solid foundation and build up.

Hate to burst your bubble, but the Autobahn doesn't last forever, either, even if it might be better made than the roads in the USA. And in Switzerland they're experimenting with special asphalt with even longer durability (IIRC with nano particles in it), as closing the highways crossing the alps has an extreme economic impact... Also, there are lots of bridges in need of maintenance in Germany as well (but not on the brink of collapsing at least) - for example that bridge that held the record of the shortest construction time also held the record of the shortest time till needing repairs.

Of course I agree with you that investment into infrastructure is a better idea than investment into lottery players (aka bankers)... which is true not only in the USA.

Re:I know what you're talking about

[BlueUnderwear]

But it's not just slashdot. ALL websites are bum rushing the add more crap idea.

Correct. But as a geek site, slashdot should know better and lead by example.

And yes, other companies do look towards (perceived) geek sites such as slashdot, gnu.org and redhat.com in order to justify their own inadequacies. A while back, our company was putting a new website online, which had huge horse blinkers. When I pointed this out to the webmasters, their response was yeah, but just look your geek friends at gnu.org (which indeed had small blinkers at the time) and redhat.com (which is just fugly).

The situation has become so bad that even the pirate party has sites where half the links won't work, where the only way to make a donation is Paypal (even though most potential donators are local, and could use an IBAN bank account number).

So, slashdot, digg, heise.de, freshmeat, gnu.org, redhat: cut down on the crap, it's not only your own sites that you are littering, but the internet as a whole! Or, if you're actually enjoying turning the Internet into a landfill, then please stop the hypocrisy of posting articles complaining about it.

Re:I know what you're talking about

[HarrySquatter]

No, it had plenty of bugs before the Ajax rewrite. Like the longstanding pagination bugs that they eventually just gave up on.

Re:I know what you're talking about

[jmorris42]

> Considering the assets required to not have the game look like shit at native resolution compared to the low-res Palms, yes.

Oh you 20th Century primitives with your stacks of bitmaps at every possible resolution. Scalable art is where it is at. And it also tends to be smaller than even a single bitmap. But that gets back to my original point, so long as there isn't a price to be paid for being ignorant nobody will bother going to the trouble of changing ways that worked well originally. In the old days throwing a fixed resolution bitmap at the problem was the simple and best solution, and this was adapted to the new problem of multiple resolution displays by simply packing multiple versions of all art because it was the easy fix requiring the least change in thinking. And if it bloats, who cares; yet.

There are two schools of thought

[bigredradio]

From my own observations, there are two schools of thought.

1. People who think anything older than 6 months is ancient and obsolete.
2. People who say, "if it ain't broke don't fix it!"

Seems the former spend their time fixing things and the later spend time bitching about "damn kids" and their lawns.

Re:There are two schools of thought

[Anonymous Coward]

not really. seems the former are a bunch of fucking idiots constantly tampering with shit that doesn't need tampering with (gnome 3, anyone? unity?), while the latter are either refreshingly pragmatic, or, as you say, tedious old farts resisting progress for the sake of pretending to be smart (windows xp, anyone?)

my view is fuck the both of them, and there must be a third school of thought. however, i'm too tired and drunk to think what it might be.

Re:There are two schools of thought

[Lumpy]

and you forget those of us in the middle. we buy the 6 month old gear for $0.10 o nthe dollar off of ebay and get to use higher end gear from the used market for lower price.

No company needs 1000bt for the accounting and sales department. But there is always some moron IT guy out there that thinks they do so they scrap all their perfect 100bt gear. and I snap it up for nothign and sell it to small businesses for a profit.

Re:There are two schools of thought

[Eristone]

No company needs 1000bt for the accounting and sales department. But there is always some moron IT guy out there that thinks they do so they scrap all their perfect 100bt gear. and I snap it up for nothign and sell it to small businesses for a profit.

I see you aren't using more recent accounting and CRM/ERP packages and don't have people pushing multi-megabyte PowerPoint and video presentations around. (or in my case - Sales pushing around vm images of a couple gig) Or people moving between desks from other parts of the company. That moron IT guy that replaces everything with 1000bt gear is sitting there going "There. Now I don't have to worry which switch the conference rooms are plugged into or if the head of HR and the CEO snag someone's office so that person goes to an empty desk to do something... "

Re:There are two schools of thought

[gstrickler]

1000bT isn't necessarily for the bandwidth. For many environments, it's used for the reduced latency. In one particular case, we had 100bT from desktop to switch, 100bT sw2sw, and 100bT to server. Replacing the switches with a single 24x1gb with 1gb links to switches with 2x1gb + 48x100mb, then 100mb to desktop more than doubled performance of one critical app. In most instances, bandwidth wasn't a factor, but the reduced latency can be of tremendous benefit for some apps, including many accounting apps.

But I agree with the rest of your post. Stay off the bleeding edge. Be selective about buying and you can get high performance machines that last you 4-8 years, maybe a bit longer, and never pay top dollar for them (even if you're buying new). The secret is knowing what you need, then buying a quality machine that's expandable/upgradeable and is slightly more than you need now.

Re:There are two schools of thought

[HapSlappy_2222]

My lawn's obsolete! Fix it, you damn kids!

Re:There are two schools of thought

[rev0lt]

Most "modern" (3 year old and newer) machines do have Gigabit connectors, so why not use them? On local networks, there are several advantages:

1) reduced latency (someone else has mentioned it) - it really helps a lot some applications;
2) less time loading roaming profiles / less time spent refreshing network shares;
3) increased bandwidth (even at 100Mbit) - Gigabit gear is usually more error-resistant, and implement smarter and faster error correction;
4) inter-departament high-speed sychronization - good for replicating storage, machine snapshotting/CDP, distributed filesystems and such;
5) instant 10x speed upgrade on recent infrastructure, since 1000T is Cat5-based (no scrapping except the switches)

My internet connection alone has 120Mbps downstream. And yes, I use it.

How things change, how they stay the same

[cusco]

There was a time when my 486/25 with the 120 megabyte hard drive and the 14.4 modem was "all you'll ever need". That didn't last long . . .

Re:How things change, how they stay the same

[DarkOx]

Its still all you ever needed its just not all you'd ever want.

Re:How things change, how they stay the same

[WillAffleckUW]

I remember 110 baud. And soldering your own circuit boards for S-100 computers and tuning your drives with an oscilliscope.

Not to mention slide rules. Not the plastic kind - a fine grained wood one.

Your mileage is not my mileage

[WillAffleckUW]

As we in the military, research university, and government spheres move to IPv6 and Internet So Fast It Makes Your Ears Bleed (tm), have you ever considered that perhaps it might be slow for you but not for us?

I mean 1000 Gbps is considered normal here, and some of us are running on faster connections, using less energy total to do the same thing.

We rarely print things anymore, and just because you have slower access to resources, you have to realize it could be because, in the war between Urban America and the rest of the country, Urban America with its more efficient energy usage and lower distances traveled - basically won the war.

Re:Your mileage is not my mileage

[ColdWetDog]

... you have to realize it could be because, in the war between Urban America and the rest of the country, Urban America with its more efficient energy usage and lower distances traveled - basically won the war.

Good. Then you can eat all the Internet you want. We'll keep the food.

Yours Truly,

Rural America

(I'd expand this comment but it takes a long time to get stuff uploaded on our 300 baud lines.)

Re:Your mileage is not my mileage

[similar_name]

I have to ask at 1000 Gbps are your hard drives even able to write that fast ? That's 125 Gigabytes per second, 500 MB/s is pretty good for an SSD. Also, what are you doing that requires that kind of speed?

Re:Your mileage is not my mileage

[Lumpy]

"Urban America with its more efficient energy usage and lower distances traveled - basically won the war."

Until the power goes out. then I own you with my farm and it's source of food you dont have.

rural america will always rule urban america. You cant raise cows in central park.

Re:Your mileage is not my mileage

[wmbetts]

Good luck feeding a country like that.

Re:Your mileage is not my mileage

[hawguy]

I have to ask at 1000 Gbps are your hard drives even able to write that fast ? That's 125 Gigabytes per second, 500 MB/s is pretty good for an SSD. Also, what are you doing that requires that kind of speed?

We have 8 blade servers with SSDs, each blade keeps most data in DDR3.

What are we doing? Medical and statistical research. You should see some of the protein folding units.

DDR3-1600 will give you a peak transfer rate of around 13GB/second. You can get higher throughput by interleaving across banks, but the Xeon 7560 (for example) will peak at around 15GB/sec

PCIe Gen-3 x8 will deliver around 8GB/second.

The fastest interconnect I've seen on a blade chassis has 10Gbit ports.

Are you sure that "1000 Gbps is considered normal here"?

Re:Your mileage is not my mileage

[Alex Belits]

Which are you supposed to be for the purpose of this thread, my karma-whoring, Microsoft-astroturfing friend hairyfeet?

Re:Your mileage is not my mileage

[garaged]

Oh yeah, like five tomatoes worth for every person every year, I can see america's overweight problem disapearing soon :)

Re:Your mileage is not my mileage

[pla]

If Urban America ever stopped providing tax money to Rural America

...Urban America would starve to death within a week, while...


the rioters in the farmlands would run out of gas money in minutes.

...The Farmers wouldn't notice for about a month. They buy diesel in bulk and store it in 500-5000 gallon on-site tanks. And while the foreign oil may arrive via barge to NYH and the parasites at the CME take their cut right off the top, the refineries have more to do with "rural" than "urban" as well (though more because no one wants to live near them than location).

Cities do share resources more efficiently than rural areas. But "more efficient" doesn't actually mean "self sustaining" - That farmer, while sucking 1000x as much energy per acre than the Manhattanite, wouldn't notice if NYC vanished tomorrow. The opposite doesn't hold true.

slow where

[magarity]

My work Pc is slow and has trouble connecting because of the n layers of Corp security whatnot. My home Pc is reasonably fast and always connects quickly.

Re:slow where

[jrminter]

Ding ding ding - we have a winner. Our IT folks put so much crapware on our corporate image, that I had to take all my lab computers out of the domain and run vanilla installs w/ minimal antivirus and our imaging hardware/software. Makes a BIG difference.

Lots more than just CPU and transfer resistors...

[mlts]

IT is a lot more than just CPU and the amount of little switches on a die. Yes, those get better and continue to do so, but there are a lot of bottlenecks that are not going away anytime soon. Until these are dealt with, things will stay almost the same in the IT world.

Couple examples:

1: Wireless bandwidth fees. This has gotten worse as time progresses. Two years ago, my T-Mobile CLIQ had unlimited tethering. Now, if I want to transfer 500 gigs of data, I'd have to pay my provider over five digits for that month.

2: Regular bandwidth. A year ago, bandwidth might be throttled on P2P downloads. Now it is metered as well on most ISPs.

3: Backups. The enterprise has the advantage that once they pay for the LTO-5 tape drives, individual cartridges are cheap, rugged, and have a lifetime guarentee. Individuals usually don't have the cash for the drive, so have to deal with hard disks which usually have a year warranty, and there is no consumer level software to handle backups, where it knows where a specific revision of a file is on what volume, be it a primary volume, or a copy saved in a safe deposit box somewhere. The enterprise has NetBackup, TSM, Networker, and other items. So, there is a major issue with making sure data is saved safely for anyone who can't afford to stick an EMC VNX array in their garage.

In the past, tape drives were not just affordable by consumers, and kept up with hard disks, but usually had some decent software that could help find media in case of a disaster. These days, there are not any good consumer level backup utilities, especially ones that can restore bare-metal.

4: Encryption. As grows storage grows the need to protect the data from everything from tapes falling off the pickup truck to hard disk drives getting yanked out of arrays.

Just raw CPU power may help things, but that is more incremental than anything else. Right now, IT is more affected by the BYOD trend than it would be by any CPU revolution. What would stir the pot would be bandwidth increases that don't have corresponding fee hikes. Having the ability to have fiber-channel bandwidth over the WAN fabric on the cheap would revolutionize things.

Two no three major flaws with this.

[Anonymous Coward]

1. Computer hardware is not a finite resource like coal is or any other natural resource. Prices go up; somebody build a plant to make more. Econ 101.

2. This assumes that computer hardware will be used the same way as it has been in the past. We are already seeing major changes. Less individual storage and more online storage; different devices that are less hardware intensive and computing is being used differently - less desktop and more handheld and all the differences down the chain from that.

3. No mention of significant technology changes. Who's to say will still be using the current architectures or even silicon tech in the future. This assumes the same old same old for the future.

This seems familiar

[RussR42]

Don't we hear this same story every so often? Before it was trace width or storage density or whatever. Perhaps some day we'll run out of tricks to making better cheaper hardware but there seems to be a long way to go yet. I mean, we don't even have tenth generation AI hologrammatic computers with IQs of 6,000 yet!

International IT expendings

[gmuslera]

will rise a lot when they have to move to local servers and companies to avoid the intrusion on their private data mandated by US government

That was the stupidest thing I've read in a while.

[Brannon]

I'm not even sure where to start other than to say--technology is only ever adopted broadly if it is cost-effective to do so. The printing press wasn't successful because of some incontrovertible march of progress--it was successful because it was cheaper to make books that way than by having monks transcribe them by hand. Yes, that caused more people to read which drove up the demand for books. And I'm sure some jackass back then wrote an article saying that demand for books was accelerating at a rate that we weren't going to be able to afford enough printing presses anymore.

Human perception

[Dan East]

There are limits to what will be demanded, and we have reached them in some areas already. Audio is a good example of this. The storage and bandwidth requirements for good (as in good enough for 99% of the population) audio is now a very small drop in the bucket. How many songs can you fit on a 16 GB micro SD card the size of your fingernail? How many songs can you stream real-time at once on a typical broadband connection? We have surpassed the technical requirements for audio by such a massive margin that it isn't even a consideration when purchasing hardware or bandwidth.

There are limits to video too. These so-called "retina" displays are a good example of the resolution limit of the human eye (we passed the color depth perception limit a good decade ago). The eye cannot discern individual pixels within the normal focal range (by the time you bring it close enough to the eye to make out individual pixels, the eye can no longer keep it in focus). We have a long ways to go to be able to store and stream video at such high resolution. However we will reach it before too long. Then it's a matter of how many hours / days of video do you need to store on how small of a device, and how many video streams do you need at one time over your internet connection.

One day we'll be moving and storing movie-length retina-resolution video with the same flippant ease as MP3s today. When we've reached that point, what would we need more bandwidth and storage for? Not for anything by human consumption - and that is the key factor.

IT spending dropping dramatically

[DogDude]

In our company, IT spending is actually dropping, even as we expand. The cost of used hardware is insanely low because of all of the individuals and companies who still feel the need to buy "new" equipment so rapidly. We have no problems running Pentium 4's and Windows XP throughout our business, and wil do so for the foreseeable future.. We've moved our email/backup/web hosting services out to providers, and all of that is sill insanely cheap. Tech has actually exceeded our needs, so our IT spending has dropped significantly. Keep buying new machines every few years, people! We're loving buying your completely functional equipment at yard sale prices!

Re:IT spending dropping dramatically

[jrminter]

I agree. I was recently pulled into a project to develop some software that was going to run on a system with a highly-customized real-time Linux kernel built from scratch from the 2009 version of Ubuntu (Karmic Koala.) I needed to make sure my code ran on that platform, so I grabbed an old (2007 vintage) laptop and installed Karmic. I was surprised how peppy it was. I suspect that it would do 99% of what most students and office workers would need. The problem is that designers keep putting out content that use new versions of Flash and other plug-ins and I suspect that these kind of annoyances are what will force people to upgrade otherwise fully functional systems. Note that vendors do this to force upgrading to new hardware and software to drive sales, not because of true need by customers. But that IS how the world works...

Re:IT spending dropping dramatically

[AbRASiON]

You can buy very very reasonable used Intel Core 2 Duo / 3 year old IBM / HP / Dell workstations, add a 100$ SSD and 50$ of ram and have a PC which performs as fast or faster than the 800$ new boxes for literally half or less of the price.

I wouldn't recommend it for a large company but mid size it seems quite reasonable to me.

A hurricane will save us

[msobkow]

When you lash together the disparate clouds of application, compute, and storage facilities from the various vendors in that space, and truly begin to tie them together as distributed applications, an amazing thing happens.

The work load distributes. The storage requirements distribute. The compute requirements distribute.

And the more distributed they become, the closer we approach a true peer-to-peer architecture.

Now take it one step further, with each person having their own "data server" nodes in their home or leased from such cloud providers. Your device is no longer used for storage, but just presentation. It caches the data from your server(s), but it doesn't need to keep the data unless you expect to use it again in the near future. Your whole SSD/HDD system in the device becomes a cache, similar to the Andrew File System, but using different communications technologies including torrents that map into a virtual file space, and private downloads directly from your data servers for content that you own personally.

Suddenly you realize the problem is not that we need infinite capacity, but that we need to break the mindset that industries like banks "own" the data. They don't. It's OUR data, and it should be on OUR servers, with them needing OUR permission to access or modify it.

Problem solved.

Peak Computing?

[slew]

If I gather what this article is speculating on, it's a phenomena similar to peak-oil.

Peak-oil doesn't necessarily mean that you run-out of oil, it just means that the marginal cost of producing more oil reaches a point which causes the rate of oil production to decrease. In the backdrop of increasing demand, and limited supply this implies a sharp downturn in availability of oil at historical prices.

If applied to computing, it would imply a limit to computing resources. I don't think we are there (although computing takes lots of electrical power and there seems to be enough semiconductor manufacturing capacity for the moment), but we may be at a point where demand increases beyond the rate at which technology can keep it on its historical increasing MIP/$ trend. If this MIP/$ trend flattens out, it may be difficult to find funding for new technological advances and fundamentally change the market for computing.

Re:Peak Computing?

[SplashMyBandit]

Nice explanation. At the moment I think the actual physical computing growth is fairly easily covered since it is a fixed cost and quite cheap for capital expenditure. What is very expensive and doesn't scale well is software licensing. I've been on plenty of projects where all resources were available apart from the money for expensive licensing (try getting LPARS off a third-party provider for a dev, integration and production environments, then get enough for Internet scale; or pay for Oracle or Enterprise SQL-Server licensing for thousands of machines, ouch). This is one area where Linux shines economically - licensing for Internet scale business (just add hardware, zero cost for the software, and only a few *good* admins needed).

Bloated apps.

[toonces33]

It isn't so much that users are expecting more from the apps, but that application vendors bloat their software as time goes on so that newer versions really only run on newer and faster hardware. I won't point fingers too much - there are many offenders here.

And on top of that, the industry is using more Java which is as slow as snot. The attitude seems to be that if it runs slow, then throw some more iron at it.

I remember my first Linux box - i486 at about 90MHz. Those were the days..

Re:Bloated apps.

[Anonymous Coward]

the industry is using more Java which is as slow as snot. The attitude seems to be that if it runs slow, then throw some more iron at it.

It's time this myth was debunked.

Java itself is not slow. Properly written and optimized Java code runs almost as fast as equivalent C/C++. (I know, I write such code, and I measure timings for operations in nanoseconds in Java.) The JIT compilers built into modern JVMs generate very optimized machine code. (I know, I've looked at the assembler output.)

Unfortunately, Java has a tendency to magnify poor programming decisions, and it's easy to be an idiot and still write Java "code." Sources of poor Java performance are generally instantiating way too many objects per second, resulting in frequent garbage collector use, and poor choices of algorithms and data structures for the underlying problem. (The same thing could be said of C/C++, although with manual memory management in C/C++ you're far less likely to run into a problem allocating/freeing too much memory per second.)

Personally, I'm fine with this as there will always be career opportunities in rewriting/optimizing someone else's crappy code.

hello self licking ice cream cone

[Dolphinzilla]

I read the headline for this story and laughed - it doesn't matter how much faster my computers or networks get - Our IT department just installs more and more virus scanners, software maintenance tools, firewalls, monitoring tools ,etc.... Each computer I get has more CPU cores and memory and faster graphics and they are able to do less and less and take longer and longer to boot. I figure before too long I'll have to go back to my old TI-30 calculator and some engineering graph paper and I'll be equal in efficiency to my computer once I factor in all the time I spend waiting for it to get around to sparing .5% of the 12 CPU cores to run the actual software I need to use....

Re:hello self licking ice cream cone

[LinuxIsGarbage]

I read the headline for this story and laughed - it doesn't matter how much faster my computers or networks get - Our IT department just installs more and more virus scanners, software maintenance tools, firewalls, monitoring tools ,etc.... Each computer I get has more CPU cores and memory and faster graphics and they are able to do less and less and take longer and longer to boot. I figure before too long I'll have to go back to my old TI-30 calculator and some engineering graph paper and I'll be equal in efficiency to my computer once I factor in all the time I spend waiting for it to get around to sparing .5% of the 12 CPU cores to run the actual software I need to use....

My work computer is a new Corei5, 4GB RAM, running a 10 year old OS (XP). The thing should fly. Yet it's much slower than my home PC which is a 4 year old AMD Dual core, 2GB RAM, running Windows 7.

A few years ago I bought a surplus PIII from work. At work with XP these machines would crawl (5+ minutes till usable). When I got it home and loaded a clean XP install, the machine flew (relatively speaking).

I'm amazed at how much crap IT departments manage to put on computers to slow them down. And how all software assumes that the first thing the users wants to do when they boot their computer is:
-Update everything at once as quick as possible
-Scan everything while that's going on as quick as possible.

And not do something like check their email, and let the other things trickle away in the background.

Another company I know migrated from XP to Windows 7. There are a lot of older computers (4 years) and everything crawls under the new IT-bloated image.

Zoning code

[tepples]

Urban farming is enjoying a renaissance.

Good luck getting zoning boards in all cities to agree to allow urban farming. Some cities have even been waging war on vegetable gardens.

FUD

[brillow]

It's all FUD. There is no reason to believe any limit is being approached. If we need more network capacity, it will be built.

The WPA?

[PeanutButterBreath]

p>

Personally I think it is high time we use an old solution to fix a new problem...bring back the WPA. a lot of our bandwidth problems would disappear if we had nationwide FTTH or at least fiber to the neighborhood. It seems like a great way to put all those sitting at home on unemployment to work and you build it right and just as many bridges built by the WPA in rural areas still work fine so too could a well built fiber network last us for ages.

I think there is plenty of old-school WPA-type work that those people could be doing. lt won't happen because it means "Big Government" giving opportunity to poor people, and that is somehow un-American.

Does not matter.

[khasim]

I see you aren't using more recent accounting and CRM/ERP packages and don't have people pushing multi-megabyte PowerPoint and video presentations around.

Does not matter. Because once it hits the VoIP with PoE for their phones it will be knocked down to 100Mb/s anyway.

gig-switch ---- VoIP-phone-with-PoE --- computer
Means that the computer is only going to get 100Mb/s.

You want to run 2x as many lines as you need to so some people can get gig to the desktop? As long as someone above you is willing to sign off on the expenses and maintenance contracts.

And I'll still be spec'ing 100Mb/s switches with PoE for the phones.

Unless you're wearing a hoodie and buying Skittles

[apparently]

and then it's still open season in the land of the free.

Stop watching the news media reporting on crime 3 states away and realize that urban violence and murder rates are at historic lows in the cities of America.

Please accept my sincere my thanks for your service to our country that has protected us from imaginary foreign threats while we're still stuck dealing with domestic threats that refuse to accept losing a civil war. I am envious of your bandwidth.

Re:Lots more than just CPU and transfer resistors.

[HockeyPuck]

Backups. The enterprise has the advantage that once they pay for the LTO-5 tape drives, individual cartridges are cheap, rugged, and have a lifetime guarantee.

Who modded this moron up? Obviously, he's never had to buy LTO-5 tape drives in bulk. I don't mean a few boxes totally 25 tapes, but hundreds and THOUSANDS of cartridges. LTO-5 isn't cheap. The enterprises may be upgrading their tape drives, but the cartridges that are often bought are LTO-4 because they are so much cheaper. Plus they can still be used in the LTO-4 drives, for which putting an LTO-5 media in an LTO-4 is a waste of $$.

This is why backup to disk is moving in. Media is expensive and restore times are slow. However, backing up to tape is actually quite fast. Still requires a pretty fast source and server to saturate an LTO-5 drive with compression.

Having the ability to have fiber-channel bandwidth over the WAN fabric on the cheap would revolutionize things.

Hello. FC over WAN is called FCIP. It's already here and used as an ISL (inter switch link) rather than a host to array (or array to array) method. Yes I know that EMC has 1GbE blades for their arrays which do replicate SCSI over IP, but that's a proprietary solution and doesn't scale. How do all the arrays move data from site A to site B? They plug into a switch via FC and then the switch encapsulates the FC packets in IP packets. Then it's up to good old TCPIP to get you to your destination.

FCIP isn't exactly cheap, but the alternative, FC over DWDM is quite a bit more expensive. Go price out your next Cisco 15454 and get back to us.

Rest Assured

[tunapez]

What Intel giveth, Microsoft taketh away.

Re:Backup and your cap

[pla]

Set a bandwidth quota at the router?

Brilliant! If I just tell the router not to go over my 10GB monthly cap, I won't go over my monthly cap!

But... Wait... What if someone actually produces more than 10GB/month of data?


manage your data at the source better?

Ah! So people just need to do less, brilliant! They could... Sleep more! Or take up solitaire. Or Knitting. Those damned kids with cancer can just wait, the next Einstein of biomedical research needs to take a nap while his bandwidth cap recovers.


Allocate resources as needed?

What, precisely, does that actually mean?


This isnt rocket science. You have a specific amount of X to use, meter accordingly.

True, and absolutely false. We have an effectively infinite bandwidth - Our ISPs would rather book double-digit profit margins this quarter than actually do something so mundane as lay another fiber, however.

Meanwhile, South Korea, a friggin' 2.5th-world country, has ubiquitous FTTP.

I've had to do desktop upgrades due to that

[dbIII]

I agree, and here's an anecdote for what it's worth. One PC was working perfectly for all work related tasks - and then the user started mucking about on Facebook and the web browser brought the system to a crawl. That was a while back but still it's insane to hit hardware limits and get to 100% CPU for a few seconds just to put a single page of text and a few pictures on a screen. Even if it's got a good reason to take a while at least give the user something to look at in the meantime.

Better Idiots?

[Taco Cowboy]

I always suspect I'd be learning something new whenever I visit /.

Thank you for proving it

Old News.

[DaneM]

I'm not sure if anyone else has taken note of this, but this paradox has been seriously re-visited concerning modern technology in the fairly recent past, near the beginning of the rise of computers in the 20th century.

When sewing machines, vacuum cleaners, toasters, microwave ovens, frozen meals, etc. became popular and readily available in the 1950s, the overwhelming assumption was that homemaking women would suddenly find themselves bored, and with nothing to do. The assumption was that the great efficiency with which housekeeping tasks could now be done would mean that the women would do what they always did in 1/2 the time, then spend the rest of the time without anything to do. What actually happened, however, was exactly the opposite.

Suddenly, the level of housekeeping that was expected went way up! The homemakers didn't do their normal duties and then stop; they made the homes more spotless and perfect than had ever been practical before, thanks to these wonderful new technologies. Now, instead of being bored, homemakers were running themselves ragged (partly at the behest of others and the new culture of perfection).

This is exactly what's been happening to everyone over the past 20 years or so, thanks to computing. It makes us work harder and faster, and runs us ragged. Likewise, increased efficiency means increased demand (since my work will undoubtedly require support from some other person, business, or industry--such as webhosting or tech support), so for every person/business/industry that does more, somebody else also has to do more. It's not just our computer hardware that's about to hit a wall; the stamina of human beings is already nearly there. (Parts of Asia have had some major problems with this in the last 20 years, what with kids going bald from stress, workers committing suicide because they can't meet demand, etc.) I strongly suspect that this increased pace of life is also a major contributor to the increasing pandemic of mental health disease (more research required to verify).

While this technological "wall" is certainly something to take notice of, it's really just a "second fiddle" to how we, as biological creatures of limited capacity, will soon find ourselves saying (at the behest of our minds and bodies), "enough!" Personally, I look forward to that particular outbreak of common sense. While I love computers and the marvels they make possible, I absolutely hate how we've allowed our technology to rule us, instead of the other way around.

Tinfoil hat time

[Kupfernigk]

Europe is full of small, cheap, fuel efficient cars. Your problem is that there was a size and power war on American roads. While I would feel perfectly safe driving a Fiat 500 Twinair or its equivalent around most of Europe, I would be terrified driving it in the US. By the time European designs make it to the USA, they seem to be carrying around a third of a tonne of additional padding and reinforcement to protect against rednecks in light trucks or middle class mothers talking on the phone in their main battle tanks. It will take a long, long time for the USA light vehicle fleet to get down to sensible sizes.