I've worked on big IT projects, and I've worked with government people who've worked on them, or managed or procured them. One director at Livermore Labs in the late 80s commented that he'd never seen a billion-dollar computer project succeed - it's just too big to do the communications that are needed to make it work, through the requirements, design, and management parts, and he was trying to work on how to break projects down into things that were small enough that they could be managed and implemented. Even the successful things are messy at large scale.
This was long before Agile (which is pretty tasty Kool-Aid, for some kinds of projects, but has its own limitations).
This was a summer job in college at a small company, and the data entry clerk probably got paid more than I did, and I was on a later shift than she was (since only one of us could use the computer at a time anyway.) I don't care what it cost IBM to pay their tech support people. And we didn't know it would take 5 hours to fix the thing when we started, or we would have talked about whether to just have her retype it.
(I read about this on the net, so almost all of the details may be wrong.) There's some company in Norway or somewhere that has a business model of putting small datacenter units into people's homes, with a narrow rack of computers and storage, using them as electric heaters, which reduces their HVAC and real estate costs and takes advantage of fiber infrastructure.
Hydro power is anything but clean and cheap, if you look at the actual environmental effects of building the stuff and preventing the floods that many of those dams were built to prevent.
Somebody else mentioned the risk of serious earthquakes - that doesn't mean you shouldn't put a data center there, both for low operating costs and proximity to part of your customer base, but you need to look at having backup capacity in other geographical areas. (Which you need to anyway, since that data center in Oklahoma might get tornadoes, the one in Arizona might get heat waves, the one in Virginia might get climate-change flooding, the one in New Jersey might annoy the governor,....)
I keep several of them around to absorb different kinds of junk mail. One of them's for reading Flickr. Another's the contact account for the Gmail account I use for watching YouTube. Another's one I started giving vendors years ago. Another one's for reading Yahoo groups, which has something vaguely resembling my real name. I've probably forgotten a few others. And no, thanks, none of them need my Real Life Phone Number. If I forget the password for the one I read Flickr with, I can create another.
Ok, I'm from a different era, but learning pointers wasn't that hard, though it was pretty much the key to C programming. It took a bit of reading library source code to really understand what K&R was saying about them, by seeing how other people used them, but they were amazingly simple and powerful.
The problems with pointers were that you always had to be careful that they were pointing to something appropriate, and that you had to know how big the things they pointed to were (in case they were passed to you from somewhere else or you were passing them _to_ somewhere else), so you still had to always always ALWAYS be careful to either do bounds checking, or put sentinels into the data, as well as checking that a pointer that some function handed you wasn't NULL or some other evil value.
And reading library source code included looking at things like strcpy() doing while(*feet++ = *bullets++); and noticing that it was much more efficient not to bother checking that "*bullets" would hit a zero before "*feet" ran out of places to store it, because obviously the person doing the function call would have checked that, so you didn't have to worry about shooting anybody else's feet. And then you have to learn not to do that yourself, which way too many people haven't learned over the last 35 years, which is why almost nobody should be allowed to program in C, even though it's my favorite programming language.
The mainframe failed significantly twice that I knew about when I was in college. IIRC, it was a 370/168. Once was because the water-cooling plumbing leaked all over the motherboard. Another time was because it really didn't like it when they added the fourth megabyte.
The System/34 minicomputer I used for a summer job failed because we were a steel fabricating company, and somebody in the shop was looking for the circuit breaker for his welder, and turned off our power instead. That would be fine, but our data entry clerk had been typing in a list of parts for a bid for six hours, and while the OS had saved where the file started on the disk when she opened it, it wouldn't save where the file ended until she closed it. So I spent about 5 hours on the phone with tech support, walking through the equivalent of fsdb to find all the blocks of the file, store their locations in the inode-equivalent, and write the end-of-file marker on the disk.
Hey Mutant, most adult humans can't digest milk.
on
A Fresh Take On Fake Meat
·
· Score: 3, Insightful
Yeah, I called you a mutant. I'm one also, from the Northern/Western European version of the mutations that let adult humans digest lactose. (There are other groups of humans that also have that - the Masai in Kenya, for instance - and most of them evolved independently about 5000 years ago.) Most normal humans are lactose-intolerant as adults, so they get indigestion if they drink raw cow milk, though most of them can handle cheeses and some other sufficiently fermented milk products.
Theoretically I can drink milk; in practice I almost never do, unless it's got coffee or cocoa in it, or it's on cereal or something.
Some archaeological work I read in the news recently has found evidence of cooking that's about 800,000 years old, and there's some suspicion that we may have been cooking for more like 2 million years (not sure if that's meat or vegetables, but cooking root vegetables makes a huge increase in the accessibility of nutrition which was important for our evolution.)
We've also got about 10-20,000 years of agriculture, which allowed humans to go from small bands of hunter-gatherers to large civilizations, and most of that's grain agriculture, not just herding.
But just because meat's tasty and we've got senses designed to tell us that it's tasty (or when it's spoiled), that doesn't mean we have to kill animals for food. Doesn't mean I wouldn't like a veggie burger that tasted like Real Dead Animals on occasion.
My experience as a vegetarian wandering around the Germanies on vacation is that if you leave out the beef and pork, and convince them you don't eat fish or chicken either, the cuisine is, if not light-weight exactly, at least light enough to lose weight while still not being hungry all the time. Sometimes you have to resort to beer being the protein course of your meal, but that's ok.
(ObMOOOO.) Cows are usually vegetarian (not vegan, of course, but it's not exploitative for calves to drink cow's milk.)
Chickens aren't - their natural diet includes insects, but if your corpse were lying around and had time to rot a bit, they'd totally eat you too. Pigs and goats would probably wait until you're dead, but not much longer. Almost all fish eat other fish; it's only the really bottom end of their food chain that eat plants. Sheep don't have the same reputation that goats do about eating anything they can get, but I suspect their tastes are similar.
RDP is simply not an adequate substitute for a network-transparent window system. Yes, it'll let you do some things badly, and other things mediocrely, but that's about it. And I haven't seen any evidence that the Wayland folks understood that early on, so I haven't kept up with Wayland when there's working X.Org.
There was a recent crowdfunding campaign for a open-protocol switch (I forget if it was OpenNFV, OpenFlow, or OpenVSwitch? Probably NFV.) 4-port 100 Mbps, so easy enough to do cheaply these days. I didn't really have any experiments I wanted to do with one that I couldn't also do with a virtual switch, so I didn't join the crowdfunding, and for production work I'd want at least GigE, but it was still interesting thing to go by.
How long ago did 300dpi printing become obsolete? These days I usually print drafts at 600dpi, because laser printers and LANs are fast enough that it's not annoying, and I don't usually explicitly notice jaggies at 300dpi, but you can still tell that the higher resolution looks better, if you care.
But that's black and white text printed on dead trees, not screens. Sure, it's harder to notice minor resolution differences with color photographs than with letters that have well-defined edges, and even harder to tell with moving images, but if you're using anti-aliased text on your screen, because it just looks better than non-anti-aliased, that's because you need more pixels. And yes, you've got enough GPU horsepower these days to trade the processing needed for anti-aliasing against the higher screen resolution, but you're doing it because your screen resolution isn't high enough.
I'm using a 17" 1920x1080 screen, and I'd like more pixels. This is generally good enough, with anti-aliased fonts, and the 22" 1080p screen at my office looks surprisingly good, but I'd still prefer 2560 instead of 1920, and the big advantage of 4K would be to have two readable pages side-by-side, which means more pixels vertically. (Sure, 16:9's fine for watching movies, but that's very seldom what I'm using that screen real estate for.)
It was Christmas Eve, somebody lifted the Visa card out of my wife's purse while we were at dinner. They bought coffee at a mall (successful), then tried to buy a TV at a Radio Shack 10 miles away (failed), and we got a phone call from the credit card company. It wasn't my home state (visiting family, and my mom actually did need a new TV:-) Successful detection!
But I've also had a couple of rounds of false alarms, where I've been traveling somewhere and gotten the "Card declined, call us" when I tried to use the card, because their fraud detector triggered on purchases in an unusual city - even though I'd also used that card to buy the airline tickets:-) They should have done better.
The only other times I've had credit cards physically stolen were once when my wife's purse got stolen (we canceled the cards before they got used), and once decades ago, back when credit card verification was handled with little paper books, and I had to go into the Sears store in Oakland and give them 25 signature samples (which felt a lot like I should also be writing "I will not let my credit card get stolen again".) The thief, or somebody they sold the card to, eventually bought about $1300 worth of stuff over a few months, even though I'd reported the theft and I wasn't liable for any of it.
Old Apple customers aren't a drain on Apple's financials, even in between the times they're buying new shiny Apple products, but that's Apple.
If you're selling competitive-market hardware like Android phones, you not only need to sell your new phones to new customers, you have to keep the old customers happy enough that 2-3 years from now they'll consider buying a new phone from you, or at very minimum, you're going to have to keep them happy enough they're not saying Really Terrible Things about your support of the old products that trash the willingness of new suckers to buy your products.
For instance, I'm currently a semi-happy Samsung customer, though I've heard rumors they've abandoned my G4 mini. It took me about 8-10 months to go from being a happy owner of a shiny HTC phone to being a disgruntled one (the Aria had a highly customized Android 2.1, locked to the Android Market, and by the time their highly customized 2.2 came out, my phone would no longer accept any software updates, because Google Play was not the same as Android Market.) And Coby? Sure, I knew it was a low-end no-name tablet, but even the manufacturer's web site appeared to have forgotten the product's name by the time I'd opened the box, though on the other hand, Google Play keeps working just fine on it, so until 4.0.4 becomes totally unsupportable, it's doing pretty well.
Even better - GOP picks somebody exciting, like Ben Carson or (ok, it's too late to get Rick Perry, maybe Santorum?), Trump says "Bah, I can do better than that on my own", starts a "HUUUUGE Party", THEN we get Sanders. Or Lessig, with Sanders as VP.
(Disclaimer: I'm a Libertarian, so I get to pretend to be neutral between the two big-money parties, which is going to happen shortly after the Republicans clean up the corruption of the Rove/Cheney/Koch/Bush years, or at least get it down to the average Democratic level of traditional personal corruption, like having another Warren Harding instead of the Military-Industrial Complex's latest mouthpiece.)
Sure. It's technically much more powerful, and in theory there could be more things you could do with it.
In practice, it was as much fun as a Viewmaster that comes with half a dozen photo disks, and that was more than enough fun to justify the $5 and the time to build it (which was part of the fun), but there wasn't anything to get me to come back and use it again after a day or two.
It's a unicorn with multiple horns, obviously. Or it's a startup that hit multi-billion-dollar valuation, with or without any particularly good reason.
Sure, they'd have lost their autonomy, but they'd have been a multi-unicorn and cashed out instead of losing money rapidly as people figured out that Groupon isn't very useful and was more or less a fad.
You need to have at least three hardware servers, all with lots of memory and disk, so you can have a primary and backup for production use and another for IT infrastructure development. If you're doing both VMware and OpenStack, which is not a bad position to take, you really need 5, two per hypervisor plus a spare. Otherwise you're going to be spending your time keeping your Shiny New Virtualization Platform up to date, instead of spending it Virtualizing Old Non-Shiny Stuff, which is what you should be doing.
"Friends don't let CA buy their friends" - Several friends of mine had worked at Ingres, and when Computer Associates bought them, if you wanted to stay you had to sign a really aggressively pro-company agreement, with lots of non-compete and similar clauses (and I assume lower salaries.) They all quit, some of them in groups. CA got the intellectual property, but lost a lot of the intellect and corporate knowledge that gave it value.
Slashdot Mirror
I've worked on big IT projects, and I've worked with government people who've worked on them, or managed or procured them. One director at Livermore Labs in the late 80s commented that he'd never seen a billion-dollar computer project succeed - it's just too big to do the communications that are needed to make it work, through the requirements, design, and management parts, and he was trying to work on how to break projects down into things that were small enough that they could be managed and implemented. Even the successful things are messy at large scale.
This was long before Agile (which is pretty tasty Kool-Aid, for some kinds of projects, but has its own limitations).
This was a summer job in college at a small company, and the data entry clerk probably got paid more than I did, and I was on a later shift than she was (since only one of us could use the computer at a time anyway.) I don't care what it cost IBM to pay their tech support people. And we didn't know it would take 5 hours to fix the thing when we started, or we would have talked about whether to just have her retype it.
Meeting's down at the ice cream place.
(I read about this on the net, so almost all of the details may be wrong.) There's some company in Norway or somewhere that has a business model of putting small datacenter units into people's homes, with a narrow rack of computers and storage, using them as electric heaters, which reduces their HVAC and real estate costs and takes advantage of fiber infrastructure.
Hydro power is anything but clean and cheap, if you look at the actual environmental effects of building the stuff and preventing the floods that many of those dams were built to prevent.
Somebody else mentioned the risk of serious earthquakes - that doesn't mean you shouldn't put a data center there, both for low operating costs and proximity to part of your customer base, but you need to look at having backup capacity in other geographical areas. (Which you need to anyway, since that data center in Oklahoma might get tornadoes, the one in Arizona might get heat waves, the one in Virginia might get climate-change flooding, the one in New Jersey might annoy the governor, ....)
But hey, they've got lots of clouds up there.
I keep several of them around to absorb different kinds of junk mail. One of them's for reading Flickr. Another's the contact account for the Gmail account I use for watching YouTube. Another's one I started giving vendors years ago. Another one's for reading Yahoo groups, which has something vaguely resembling my real name. I've probably forgotten a few others. And no, thanks, none of them need my Real Life Phone Number. If I forget the password for the one I read Flickr with, I can create another.
Ok, I'm from a different era, but learning pointers wasn't that hard, though it was pretty much the key to C programming. It took a bit of reading library source code to really understand what K&R was saying about them, by seeing how other people used them, but they were amazingly simple and powerful.
The problems with pointers were that you always had to be careful that they were pointing to something appropriate, and that you had to know how big the things they pointed to were (in case they were passed to you from somewhere else or you were passing them _to_ somewhere else), so you still had to always always ALWAYS be careful to either do bounds checking, or put sentinels into the data, as well as checking that a pointer that some function handed you wasn't NULL or some other evil value.
And reading library source code included looking at things like strcpy() doing
while(*feet++ = *bullets++);
and noticing that it was much more efficient not to bother checking that "*bullets" would hit a zero before "*feet" ran out of places to store it, because obviously the person doing the function call would have checked that, so you didn't have to worry about shooting anybody else's feet. And then you have to learn not to do that yourself, which way too many people haven't learned over the last 35 years, which is why almost nobody should be allowed to program in C, even though it's my favorite programming language.
The mainframe failed significantly twice that I knew about when I was in college. IIRC, it was a 370/168. Once was because the water-cooling plumbing leaked all over the motherboard. Another time was because it really didn't like it when they added the fourth megabyte.
The System/34 minicomputer I used for a summer job failed because we were a steel fabricating company, and somebody in the shop was looking for the circuit breaker for his welder, and turned off our power instead. That would be fine, but our data entry clerk had been typing in a list of parts for a bid for six hours, and while the OS had saved where the file started on the disk when she opened it, it wouldn't save where the file ended until she closed it. So I spent about 5 hours on the phone with tech support, walking through the equivalent of fsdb to find all the blocks of the file, store their locations in the inode-equivalent, and write the end-of-file marker on the disk.
Yeah, I called you a mutant. I'm one also, from the Northern/Western European version of the mutations that let adult humans digest lactose. (There are other groups of humans that also have that - the Masai in Kenya, for instance - and most of them evolved independently about 5000 years ago.) Most normal humans are lactose-intolerant as adults, so they get indigestion if they drink raw cow milk, though most of them can handle cheeses and some other sufficiently fermented milk products.
Theoretically I can drink milk; in practice I almost never do, unless it's got coffee or cocoa in it, or it's on cereal or something.
Hu-fu
Some archaeological work I read in the news recently has found evidence of cooking that's about 800,000 years old, and there's some suspicion that we may have been cooking for more like 2 million years (not sure if that's meat or vegetables, but cooking root vegetables makes a huge increase in the accessibility of nutrition which was important for our evolution.)
We've also got about 10-20,000 years of agriculture, which allowed humans to go from small bands of hunter-gatherers to large civilizations, and most of that's grain agriculture, not just herding.
But just because meat's tasty and we've got senses designed to tell us that it's tasty (or when it's spoiled), that doesn't mean we have to kill animals for food. Doesn't mean I wouldn't like a veggie burger that tasted like Real Dead Animals on occasion.
My experience as a vegetarian wandering around the Germanies on vacation is that if you leave out the beef and pork, and convince them you don't eat fish or chicken either, the cuisine is, if not light-weight exactly, at least light enough to lose weight while still not being hungry all the time. Sometimes you have to resort to beer being the protein course of your meal, but that's ok.
(ObMOOOO.) Cows are usually vegetarian (not vegan, of course, but it's not exploitative for calves to drink cow's milk.)
Chickens aren't - their natural diet includes insects, but if your corpse were lying around and had time to rot a bit, they'd totally eat you too. Pigs and goats would probably wait until you're dead, but not much longer. Almost all fish eat other fish; it's only the really bottom end of their food chain that eat plants. Sheep don't have the same reputation that goats do about eating anything they can get, but I suspect their tastes are similar.
RDP is simply not an adequate substitute for a network-transparent window system. Yes, it'll let you do some things badly, and other things mediocrely, but that's about it. And I haven't seen any evidence that the Wayland folks understood that early on, so I haven't kept up with Wayland when there's working X.Org.
There was a recent crowdfunding campaign for a open-protocol switch (I forget if it was OpenNFV, OpenFlow, or OpenVSwitch? Probably NFV.)
4-port 100 Mbps, so easy enough to do cheaply these days. I didn't really have any experiments I wanted to do with one that I couldn't also do with a virtual switch, so I didn't join the crowdfunding, and for production work I'd want at least GigE, but it was still interesting thing to go by.
How long ago did 300dpi printing become obsolete? These days I usually print drafts at 600dpi, because laser printers and LANs are fast enough that it's not annoying, and I don't usually explicitly notice jaggies at 300dpi, but you can still tell that the higher resolution looks better, if you care.
But that's black and white text printed on dead trees, not screens. Sure, it's harder to notice minor resolution differences with color photographs than with letters that have well-defined edges, and even harder to tell with moving images, but if you're using anti-aliased text on your screen, because it just looks better than non-anti-aliased, that's because you need more pixels. And yes, you've got enough GPU horsepower these days to trade the processing needed for anti-aliasing against the higher screen resolution, but you're doing it because your screen resolution isn't high enough.
I'm using a 17" 1920x1080 screen, and I'd like more pixels. This is generally good enough, with anti-aliased fonts, and the 22" 1080p screen at my office looks surprisingly good, but I'd still prefer 2560 instead of 1920, and the big advantage of 4K would be to have two readable pages side-by-side, which means more pixels vertically. (Sure, 16:9's fine for watching movies, but that's very seldom what I'm using that screen real estate for.)
It was Christmas Eve, somebody lifted the Visa card out of my wife's purse while we were at dinner. They bought coffee at a mall (successful), then tried to buy a TV at a Radio Shack 10 miles away (failed), and we got a phone call from the credit card company. It wasn't my home state (visiting family, and my mom actually did need a new TV :-) Successful detection!
But I've also had a couple of rounds of false alarms, where I've been traveling somewhere and gotten the "Card declined, call us" when I tried to use the card, because their fraud detector triggered on purchases in an unusual city - even though I'd also used that card to buy the airline tickets :-) They should have done better.
The only other times I've had credit cards physically stolen were once when my wife's purse got stolen (we canceled the cards before they got used), and once decades ago, back when credit card verification was handled with little paper books, and I had to go into the Sears store in Oakland and give them 25 signature samples (which felt a lot like I should also be writing "I will not let my credit card get stolen again".) The thief, or somebody they sold the card to, eventually bought about $1300 worth of stuff over a few months, even though I'd reported the theft and I wasn't liable for any of it.
Old Apple customers aren't a drain on Apple's financials, even in between the times they're buying new shiny Apple products, but that's Apple.
If you're selling competitive-market hardware like Android phones, you not only need to sell your new phones to new customers, you have to keep the old customers happy enough that 2-3 years from now they'll consider buying a new phone from you, or at very minimum, you're going to have to keep them happy enough they're not saying Really Terrible Things about your support of the old products that trash the willingness of new suckers to buy your products.
For instance, I'm currently a semi-happy Samsung customer, though I've heard rumors they've abandoned my G4 mini. It took me about 8-10 months to go from being a happy owner of a shiny HTC phone to being a disgruntled one (the Aria had a highly customized Android 2.1, locked to the Android Market, and by the time their highly customized 2.2 came out, my phone would no longer accept any software updates, because Google Play was not the same as Android Market.) And Coby? Sure, I knew it was a low-end no-name tablet, but even the manufacturer's web site appeared to have forgotten the product's name by the time I'd opened the box, though on the other hand, Google Play keeps working just fine on it, so until 4.0.4 becomes totally unsupportable, it's doing pretty well.
Even better - GOP picks somebody exciting, like Ben Carson or (ok, it's too late to get Rick Perry, maybe Santorum?), Trump says "Bah, I can do better than that on my own", starts a "HUUUUGE Party", THEN we get Sanders. Or Lessig, with Sanders as VP.
(Disclaimer: I'm a Libertarian, so I get to pretend to be neutral between the two big-money parties, which is going to happen shortly after the Republicans clean up the corruption of the Rove/Cheney/Koch/Bush years, or at least get it down to the average Democratic level of traditional personal corruption, like having another Warren Harding instead of the Military-Industrial Complex's latest mouthpiece.)
Sure. It's technically much more powerful, and in theory there could be more things you could do with it.
In practice, it was as much fun as a Viewmaster that comes with half a dozen photo disks, and that was more than enough fun to justify the $5 and the time to build it (which was part of the fun), but there wasn't anything to get me to come back and use it again after a day or two.
It's a unicorn with multiple horns, obviously. Or it's a startup that hit multi-billion-dollar valuation, with or without any particularly good reason.
Sure, they'd have lost their autonomy, but they'd have been a multi-unicorn and cashed out instead of losing money rapidly as people figured out that Groupon isn't very useful and was more or less a fad.
You need to have at least three hardware servers, all with lots of memory and disk, so you can have a primary and backup for production use and another for IT infrastructure development. If you're doing both VMware and OpenStack, which is not a bad position to take, you really need 5, two per hypervisor plus a spare.
Otherwise you're going to be spending your time keeping your Shiny New Virtualization Platform up to date, instead of spending it Virtualizing Old Non-Shiny Stuff, which is what you should be doing.
"Friends don't let CA buy their friends" - Several friends of mine had worked at Ingres, and when Computer Associates bought them, if you wanted to stay you had to sign a really aggressively pro-company agreement, with lots of non-compete and similar clauses (and I assume lower salaries.) They all quit, some of them in groups. CA got the intellectual property, but lost a lot of the intellect and corporate knowledge that gave it value.