That's a good approach. At 8 drives a day that's 250 a month for a station that you can build for well under $1000. I'd only add that you may need to tune this based on the failure modes that you are seeing.
For example, if it's just bad spots, then you'll want to do as many reads and writes as possible. For that, the fastest thing would be a little C program that reads and writes different patterns to the raw device linearly.
On the other hand, if the failures are tied to seeks, you'll want to write to semi-random locations on the device, to force maximum seeks. Or if you see a mix of both, then your best bet might be to follow m0rph3us0's plan, perhaps tweaking it a bit to better simulate normal filesystem efficiency (and you can just do bit compares rather than md5sums if CPU is an issue).
You should also keep an eye on heat issues. The burn-in should happen at temperatures that are like what they will be in the end systems. If you pack 8 seeking drives into some cases, they'll cook. If you leave them in the open air, they might not trigger the failures you are seeing in the field. Try to match measured operating case temperature.
Oh, and don't forget to measure whether this burn-in is really helping. Take stats now, and keep tracking causes of return. It could be that the drives are sensitive to noisy power or vibration or something else that your burn-in won't catch.
I'm dubious anything can possibly put a lower bound on code quality. In any case, I've dealt with a couple of big Java projects where the code quality was significantly under the "easier to throw it away and rewrite it" threshold, below which it really doesn't matter.
I think we agree on everything important, but I want to point out a nuance of the keep/toss distinction for the crowd.
One of my clients has a code base that, no question, it would be cheaper to rewrite than to fix. With the resources available, we estimate it would take us 3-6 months, with 4 months being the most likely (i.e. 4 months, +2 or -1).
But they have do deliver some new stuff in 3 months; without it, the company will be out of business. The deadline is a real external one, tied through various business processes to Christmas. And we can't get more resources in this economic climate. So even though it's cheaper to rewrite, we can't do that. We'll just have to wrap the existing system in regression tests and refactor it as we go.
Assuming it's an actually running code base, I'd rather it be in C++.
Hmmm... I get your point. I think I'd still prefer Java, as the crappy C and C++ I've dealt with almost always problems related to stepping on memory. Since Java enforces encapsulation much more strongly, those errors disappear. Maybe it's just post-traumatic stress disorder, but never having a SEGV or a buffer overrun soothes my jangled nerves.
I would say divide it up based on how much each person contributed, and give divide a percent of the earnings equally among the people in that group.
I used to work for financial traders, and company bonuses were based upon how much each person contributed to the total profit. If you went around and asked each person what percentage of profits they were responsible, the total would have been at least 200%, and probably more like 500%. So payment by effort is likely to make everybody feel like they got gypped.
And note that as soon as you introduce a mechanism for determining level of effort, you are creating incentives that may not match what you want. As one person pointed out, paying by line of code leads to bloat. Paying based on self-evaluation rewards the arrogant. Paying based on some sort of voting about who did the most rewards those who announce their contributions most boldly.
Unless we're talking about a lot of money here, I'd encourage people to use the dough for non-cash rewards. E.g., that you get everybody together for a face-to-face meeting and party. Or you buy hardware and bandwidth that everybody gets to use. Or you give the money to a basket of charities that everybody helps pick. Or you sponsor somebody to go and present the project at a conference. Or you buy advertising space to promote the project. Or you buy CDs for a central streaming server for project members. And so on.
And if it is a lot of money, you probably have a chance at a much larger market, so it's worth starting a company and hiring a marketroid who can get enough sales that you can afford to hire people full time.
Java is currently held to be the best OO language ever by some people. (I am especially annoyed by the multiple-inheritance-is-inherently-evil part of the Java camp).
I do a lot of Java programming, and I can say with some confidence that people who think that Java is the best OO language ever need to get out more.
But I can deal with Java because I understand its basic theory. The notion is that the people who made it are so much smarter than you or I that they needed to keep what they thought of as dangerous or hard things out of the hands of mere mortals.
On one hand, this is bullshit. They may have been smart, but they weren't all that smart, and there are about a zillion places where that sticks out. Moreover, their instistence that there is only one right way to do things means that on the occasions where they guessed wrong, you have to resort to all sorts of stupid contortions.
On the other, if I have to inherit a big code base from somebody, I'd rather it be in Java than about anything else. The code is guaranteed not to be brilliant, but Java's limits also put a lower bound on how awful the code can be.
I think the biggest problem with java isn't how limited it is, but rather that it insists on staying so limited. It's clear to me that the people who make, say, Perl and Ruby actually use it to get things done, and they push the language forward when they find limits. But I suspect that the people in control of Java don't actually write code in it, or they would have fixed some of their more glaring oversights years ago.
You are not marketing. You did not go to school to be in marketing. Therefore, according to Marketing, you don't know shite from shinola. It doesn't matter if you have a doctorate, have coded C for 20 years, scored 300 on a standard IQ test, taught yourself Perl, MySQL, PHP, HTML, ASP, Java, and C, or cut your teeth on a PDP-10, you are an imbecile because you can't (or won't) sell.
Thank goodness we technical people don't feel that way about marketroids.
My suggestion is to build the applet, install it, let it run for a few weeks, and demonstrate the results at the next meeting.
That's a good strategy. I'd add two more things. First, provide them with something they want that the vendor doesn't provide. Often, what people really hate about outside vendors is that they don't have much control, so if you use a short-cycle iterative process (e.g., Extreme Programming), they may be instantly happier because they feel like they are in the driver's seat.
Second, learn how to express your objections and offerings in marketing terms. E.g., don't talk to them about right to privacy; talk to them about bad PR.
The scary part is that in the year the prediction has been up, nobody has been willing to bet against him. For those not familiar with the site, it's a place where people can make predictions and bets about the future. The soonest allowable bet is two years from now, but most go out a lot farther than that. So if you're so sure about your opinions on biotech (or any other topic) that you're willing to throw down on the public record, you can. (All the wagers go to charity, so don't think you'll be getting rich. But if you win, you can pick the charity.)
Time to stop being a geek. I'm getting my pencils and paper back out, doing RPGs that way,
Uh, I hate to tell you, but if you're tring to stop being a geek you're on the wrong track.:-)
I'd rather be an insurance guy or something similarly boring then spending part of my life in a 4x6 cell,
Man, you must not have spent much time at an insurance company. 40 years at an insurance company is roughly the same number of hours you'd spend in a 10-year prison sentence, and the cells are about the same size. And at an insurance company, nobody ever gets time off for good behavior.
Would there be benefits to managing 20 1048 CPU single-OS systems versus 54,000 linux machines?
Some, but they don't outweigh the costs.
A 2 GB 2 CPU box is well under $1000, right? so 32 of those would be, say, $25,000? But the SGI Altix 3700 configured with just 64 processors and 64 GB of RAM costs a cool million dollars.
Even if the multi-box solution demands more fuss to manage (higher failure rates, I'd guess), since you've saved $975,000, you can afford a little admin time. Plus, the cost of somebody who can build intel boxes must be less than half of somebody who is blessed to work on high-end SGI supercomputers. And I'm sure the SGI service and support contract in't cheap, either.
And the nice thing about a distributed approach like this is that if any given box goes down, so what? Whereas if one of your bazillion-dollar 1048-CPU boxes coughs up a hairball, that's a noticeable percentage of your regional computing capacity.
There are some applications where you can't go for this federated approach, but not as many as people think.
Yep. A single rack, unpowered, at an adequate colo in the bay area is circa $300 per month. Assuming these are 2U boxes, that's about $170 per year in space for a box.
For dual-processor, multi-disk boxes, the power draw is more than the 60-90 watts; I recently measured a bunch and they ranged from 100-220 watts. (I was impressed that this depends on CPU load; my dual-processor P3 box uses an extra ~10 watts per processor at 100% CPU, even with the disks quiescent.) Anyhow, a 20A circuit at the same colos runs for around $300 / month as well; figure two of those per cabinet and that's another $340 a year per box.
Now I'm sure Google gets the volume discounts, but even at 50% off, it's at least worth it to pay somebody to go turn off broken machines. And even if they just throw out the hardware, pulling should be worth the money, too. So hey, if the Google engineers are reading this, let me just say that I'm glad to come by and pull all of the broken machines out for you at very reasonable rates, and I'll even haul them away for free.:-)
When you offer your trained expert opinion to an idiot, expect nothing less that idiocy in return.
Many technical consultants don't realize that "technical" isn't the only word in their job description. The way I look at it, it's not just my job to give good advice, it's to do so in a way that they understand and follow.
That's not to say your pal wasn't dealing with somebody certifiable. But I'd say only 30-40% of my brainpower when consulting goes into the technical stuff; the rest is spent on understanding and dealing with people, in all their falliable glory.
Back in college I did tech support. We would frequently get calls from people who had monitors with Trinitron tubes saying that two thin black horizontal lines had suddenly appeared after months of ownership.
It took a lot to persuade them that those lines had always been there and that they had just noticed them. Some people would insist on coming in to look at identical monitors just to make sure we weren't lying to them.
That's a good list. The only thing I'd add is to suggest using something like LIDS. The basic notion is that you add something to the kernel so that root is no longer a trusted account. Instead, you give very specific permissions to each program that needs them.
That way even if somebody does compromise, say, named, they won't be able to screw with much. And since LIDS logs attempts to violate security, as soon as they try to, say, view the password file or alter the logs, I'll know about it.
I'll admit that I find spam as annoying as the next guy, but I try to take precautions and use a fairly effective junk mail filter [apple.com], so spam at best rises to the level of minor annoyance. Aren't there better things to spend our money and time lobbying lawmakers and dragging people to court about?
This sounds reasonable. People said it when spam was just getting started. They said it when spam was a few percent of email. And again when it was 10%. Now it's circa 50% and still growing rapidly.
There's no particular reason to believe that this trend will improve. And spam is a classic evolutionary arms race: every advance made in spam defense is met with an advance in spammer offense. And as with any parasite/host relationship, the parasites have a much stronger interest in succeeding (because it's their whole livelihood) than the hosts do (because spam is just one of many problems in their lives).
So our options are A) try to fix it now, or B) wait until it reaches some increased level of annoyance. Personally, I think the billions of dollars wasted on spam each year are enough to justify the effort of extermination.
Oh, great, not only would the moron be breathing down my neck while I was typing, he'd be changing my code to!
Well, if he's changing your code to exclamation points, you need a better partner.
Seriously, pair programming doesn't mean that somebody will be breathing down your neck or changing your code, and the suggestion in the post is stilly.
Real pair programming is a lot like the relationship between a driver and a navigator on a road trip. Back-seat drivers are a pain in the ass, but a good navigator can keep you on the optimal path.
ignoring my revulsion to the idiocy that is Extreme Programming.
Works for me! And a lot of others. I'm sorry to hear it doesn't work for you. Oh, wait, you have tried it before trashing it, right?
And premature optimization has been proven to be very, very bad time and again!
Agreed. And XP is in full agreement with that, too.
Instead of the CEO of McDonalds making 48zillion, he could suck it up, tighten his belt, buy US raised cattle and only make 40zillion. *gasp* a 8 zillion dollar loss. But hey - when you've already got a few billion - what's a zillion? (yes - I know... zillion... I'm trying to make a point)
Why don't you go out and try this will real numbers? Look up the McDonalds CEO's salary, calculate how many cows they buy, and how much difference him giving up his salary would make?
I'd bet the difference is much smaller than you'd think.
In Japan, a corporate exec's salary is based upon the success (or lack of success) of a company. That is a proper business model. Maybe it would make our business leaders more honest (read: ENRON, etc...).
You should learn a little more about the topic. Simpleminded solutions about making corporate leaders more responsible are one of the major causes of the recent accounting frauds.
Why? Well, it used to be that CEOs were mainly paid via salary. But it was felt that CEOs weren't responsible enough that way, that they tended to do things that were interesting for them, but not necessarily beneficial to customers or shareholders.
So some bright people said, "Gosh, let's make it so that their interested are better aligned with shareholders. We'll give them stock, and stock options. And bonuses, too."
Which in many cases worked. But it also gave a very tempting opportunity to a lot of feckless MBA-fueled locusts, who took the options, bent the corporate numbers to look better than they were, took their fat checks, sold their options, and ran. And you know the story from there.
The lesson? Yes, it's important to make people responsible, especially the executives. But simple, obvious solutions generally come with unintended consequences.
One of the reasons we thrived after WWII was due to the fact that we were a nation that produced goods for the entire world. Every part in those goods were made in the USA, by Americans.
If we made goods for the entire world, what would we get back? The answer has to be "other valuable stuff," otherwise we'll stop doing it. That's trade, and tariffs harm it.
And if we make goods only for ourselves, we end up poorer than we would if we traded. Honest. Go do some research on "comparative advantage", the economic principle that underlies trade. Or imagine what would happen if each state, each city started putting up tariffs to "protect" local industries. A giant clusterfuck, that's what.
.....didn't you just read in the posts above that now it takes both husband and wife working.....what effect do you think that has on household income (when both husband and wife work). That is the reason for the increase in household income....
It would gratify me immensely if people would go off and actually do a little research before replying. If nothing else, it would be more fun.
If you had, you would have discovered that the median male individual income has increased by about 10%, and the median female income has increased by 90%. I couldn't find median figures together, but US per capita income has gone up 93% since 1968. And all of these numbers are inflation-adjusted, but using CPI, which overstates inflation; actual purchasing power is better still.
And of course, this is with a substantial increase in population. If trade really were destroying jobs, then we would expect there to be huge unemployment, rather than the huge increases in jobs and the standard of living.
As to husbands and wives working, I note that this partly because, gosh, both men and women want to work, and there's much less discrimination against women these days. I know very few stay-at-home moms, and every one of them thinks of it as pretty temporary.
But another important factor is that we have higher expectations. Go look up the number of cars per person in 1968, for example. Or the size of the average home. Or the number of televisions. Most people also feel like they must have about a zillion consumer goods and services, many of which didn't exist or weren't widely available (microwaves, cable tv, computers, mobile phones, personal stereos, car stereos, video games, home gyms, etc, etc, etc).
So for those who are comparing with the golden ages now seen only on Nick at Nite, perhaps if you try living a similar lifestyle, you'll see how much things have improved.
Your stats about the trade increase since the 60s are irrelivant. You forget to add other variables such as inflation.
No, I don't. Take a look at US household income statistics. Adjusted for inflation, the median household income has increased 26% over the same period. And that's using the CPI, which if anything understates things. E.g., consider what kind of computing power (or home entertainment gear, or about eight zillion other things) you could have bought in 1968.
Imagine how much more the dollar would be worth if the clothing companies moved production back to the US, and gave those jobs back to Americans.
If all clothing were made in the US, then there would be a few more US jobs, true at least for a while. But to benefit a very few workers, all US consumers would pay more for their clothing, making the consumer's dollar worth less, not more. And if people are forced to spend more money on clothes, that's less that they spend elsewhere, putting other Americans out of work.
It doesn't take a genius to figure out that if you create jobs, you lower unemployment. if you lower unemployment and create jobs, people have money. If people have money - they spend it. If people spend money - it's good for the economy. *duh*
As H. L. Mencken says, "For every complex problem there is an answer that is clear, simple, and wrong."
It's true that creating jobs is good. But trade barriers don't create jobs; they just move them around. And since they move them from more efficient producers to less efficient ones, they destroy value, making everybody poorer.
"Bzzzt! Actually, there's strong statistical evidence that trade barriers make the people protected by them poorer. America historically has been a leader in lowering trade barriers, and that's part of why we're so rich."
Any proof of this? If anything its the other way around.
Sure. See this paper. Here's part of the abstract:
I use nine alternative indexes of trade policy to investigate whether the evidence supports the view that, with other things given, TFP growth is faster in more open economies. The regressions reported here are robust to the use of openness indicator, estimation technique, time period and functional form, and suggest that more open countries have indeed experienced faster productivity growth.
Although I'm not an economist, my understanding is that the correlation is pretty well established, both theoretically and empirically.
Are they really better or is it because your boss is greedy?
The basic purpose of capitalism is to make greedy people useful to society. Overall, it has worked pretty well. People complain about your namesake, Bill Gates, but I'd rather have him making billions than running mafias or starting land wars, the historically typical pursuit of the ultra-ambitious.
But I don't think people are outsourcing to India because on some absolute scale the Indian programmers are better. I think they do it because they are a better value. To keep our high salaries, we will have to find new ways to earn them.
Part of the reason for the huge slashes is that programmers are viewed as standard commodities that do not bring money into the buttom line. They do not give a shit about your skills if you do not make them money. Therfore an underskilled Indian would be more appropriate then you for the job.
So the interesting question is, are the people who think that correct? If the answer is yes, then programmers need to step up to the plate and start providing business value. And if the answer is no, that programmers do provide plenty of value, then we need to make sure that people understand that.
Also minimum wage laws helped the us economy spreading more wealth back.
There's a lot of disagreement among economists about the effect of minimum wages, but they are generally thought to be net mildly harmful to the economy. If you want to spread wealth, tax policy is a better way to do that.
A much better approach is to find a way to justify the higher wages of American works. In order to not get outsourced, find a way to prove you're worth the money. Is that sheer code productivity, product insights, or what?
My favorite is to make use of short-cycle, agile development methods like Extreme Programming. If you take full advantage of the high-bandwidth, low-latency medium of physical presence, you can achieve things that outsourcing can't match.
generally speaking "free trade" is often just code for "legal loophole"
Sometimes that's true, but it's mainly in the short term.
Take a look at what's happening to formerly low-cost producers like Japan and South Korea, the scary monsters from decades gone by. We all got richer through trade. And rich people don't like to work at crappy jobs, so they are exporting a lot of the low-cost manufacturing they used to do to countries that are still poor. Along the way, South Korea has picked up a lot in the way of democracy and human rights.
totalitarian security
This strikes me as an especially pernicious issue. My first instict is that we shouldn't do business with oppressive regimes. But on the other had, trade with China has been a big moderating influence on them, and the new generation of capitalists has been forcing political reforms that seemed impossible 20 years ago.
I wish I had a better answer than "take a middle path" between economic isolationism and lassez-faire globalization, but it seems that economic wealth and democracy (and all the regard for human rights it entails) are tightly intertwined.
By removing jobs from the US, they're hurting the economy.
Are they?
Trade has increased drastically over the decades, both in absolute terms and as a percentage of GDP (from under 10% in the late 60s to over 25% now). So by your theory, the economy should be a wreck.
But curiously, we're much better off by almost any measure. We're in the middle of a downturn, but unemployment is still under 6%, and the economy's still growing.
Slashdot Mirror
That's a good approach. At 8 drives a day that's 250 a month for a station that you can build for well under $1000. I'd only add that you may need to tune this based on the failure modes that you are seeing.
For example, if it's just bad spots, then you'll want to do as many reads and writes as possible. For that, the fastest thing would be a little C program that reads and writes different patterns to the raw device linearly.
On the other hand, if the failures are tied to seeks, you'll want to write to semi-random locations on the device, to force maximum seeks. Or if you see a mix of both, then your best bet might be to follow m0rph3us0's plan, perhaps tweaking it a bit to better simulate normal filesystem efficiency (and you can just do bit compares rather than md5sums if CPU is an issue).
You should also keep an eye on heat issues. The burn-in should happen at temperatures that are like what they will be in the end systems. If you pack 8 seeking drives into some cases, they'll cook. If you leave them in the open air, they might not trigger the failures you are seeing in the field. Try to match measured operating case temperature.
Oh, and don't forget to measure whether this burn-in is really helping. Take stats now, and keep tracking causes of return. It could be that the drives are sensitive to noisy power or vibration or something else that your burn-in won't catch.
I'm dubious anything can possibly put a lower bound on code quality. In any case, I've dealt with a couple of big Java projects where the code quality was significantly under the "easier to throw it away and rewrite it" threshold, below which it really doesn't matter.
I think we agree on everything important, but I want to point out a nuance of the keep/toss distinction for the crowd.
One of my clients has a code base that, no question, it would be cheaper to rewrite than to fix. With the resources available, we estimate it would take us 3-6 months, with 4 months being the most likely (i.e. 4 months, +2 or -1).
But they have do deliver some new stuff in 3 months; without it, the company will be out of business. The deadline is a real external one, tied through various business processes to Christmas. And we can't get more resources in this economic climate. So even though it's cheaper to rewrite, we can't do that. We'll just have to wrap the existing system in regression tests and refactor it as we go.
Assuming it's an actually running code base, I'd rather it be in C++.
Hmmm... I get your point. I think I'd still prefer Java, as the crappy C and C++ I've dealt with almost always problems related to stepping on memory. Since Java enforces encapsulation much more strongly, those errors disappear. Maybe it's just post-traumatic stress disorder, but never having a SEGV or a buffer overrun soothes my jangled nerves.
I would say divide it up based on how much each person contributed, and give divide a percent of the earnings equally among the people in that group.
I used to work for financial traders, and company bonuses were based upon how much each person contributed to the total profit. If you went around and asked each person what percentage of profits they were responsible, the total would have been at least 200%, and probably more like 500%. So payment by effort is likely to make everybody feel like they got gypped.
And note that as soon as you introduce a mechanism for determining level of effort, you are creating incentives that may not match what you want. As one person pointed out, paying by line of code leads to bloat. Paying based on self-evaluation rewards the arrogant. Paying based on some sort of voting about who did the most rewards those who announce their contributions most boldly.
Unless we're talking about a lot of money here, I'd encourage people to use the dough for non-cash rewards. E.g., that you get everybody together for a face-to-face meeting and party. Or you buy hardware and bandwidth that everybody gets to use. Or you give the money to a basket of charities that everybody helps pick. Or you sponsor somebody to go and present the project at a conference. Or you buy advertising space to promote the project. Or you buy CDs for a central streaming server for project members. And so on.
And if it is a lot of money, you probably have a chance at a much larger market, so it's worth starting a company and hiring a marketroid who can get enough sales that you can afford to hire people full time.
Amen! If I rewrite a section of code to make it shorter and clearer, I'd have to pay money.
Java is currently held to be the best OO language ever by some people. (I am especially annoyed by the multiple-inheritance-is-inherently-evil part of the Java camp).
I do a lot of Java programming, and I can say with some confidence that people who think that Java is the best OO language ever need to get out more.
But I can deal with Java because I understand its basic theory. The notion is that the people who made it are so much smarter than you or I that they needed to keep what they thought of as dangerous or hard things out of the hands of mere mortals.
On one hand, this is bullshit. They may have been smart, but they weren't all that smart, and there are about a zillion places where that sticks out. Moreover, their instistence that there is only one right way to do things means that on the occasions where they guessed wrong, you have to resort to all sorts of stupid contortions.
On the other, if I have to inherit a big code base from somebody, I'd rather it be in Java than about anything else. The code is guaranteed not to be brilliant, but Java's limits also put a lower bound on how awful the code can be.
I think the biggest problem with java isn't how limited it is, but rather that it insists on staying so limited. It's clear to me that the people who make, say, Perl and Ruby actually use it to get things done, and they push the language forward when they find limits. But I suspect that the people in control of Java don't actually write code in it, or they would have fixed some of their more glaring oversights years ago.
impressionist programing? It would only look like code from far away.
You mean VB?
You are not marketing. You did not go to school to be in marketing. Therefore, according to Marketing, you don't know shite from shinola. It doesn't matter if you have a doctorate, have coded C for 20 years, scored 300 on a standard IQ test, taught yourself Perl, MySQL, PHP, HTML, ASP, Java, and C, or cut your teeth on a PDP-10, you are an imbecile because you can't (or won't) sell.
Thank goodness we technical people don't feel that way about marketroids.
My suggestion is to build the applet, install it, let it run for a few weeks, and demonstrate the results at the next meeting.
That's a good strategy. I'd add two more things. First, provide them with something they want that the vendor doesn't provide. Often, what people really hate about outside vendors is that they don't have much control, so if you use a short-cycle iterative process (e.g., Extreme Programming), they may be instantly happier because they feel like they are in the driver's seat.
Second, learn how to express your objections and offerings in marketing terms. E.g., don't talk to them about right to privacy; talk to them about bad PR.
I'll take that bet! I figure chances are I'll be too dead to pay up anyway.
Sorry, mister. They're ahead of you on this one: it's strictly cash up front.
The scary part is that in the year the prediction has been up, nobody has been willing to bet against him.
For those not familiar with the site, it's a place where people can make predictions and bets about the future. The soonest allowable bet is two years from now, but most go out a lot farther than that. So if you're so sure about your opinions on biotech (or any other topic) that you're willing to throw down on the public record, you can. (All the wagers go to charity, so don't think you'll be getting rich. But if you win, you can pick the charity.)
Time to stop being a geek. I'm getting my pencils and paper back out, doing RPGs that way,
:-)
Uh, I hate to tell you, but if you're tring to stop being a geek you're on the wrong track.
I'd rather be an insurance guy or something similarly boring then spending part of my life in a 4x6 cell,
Man, you must not have spent much time at an insurance company. 40 years at an insurance company is roughly the same number of hours you'd spend in a 10-year prison sentence, and the cells are about the same size. And at an insurance company, nobody ever gets time off for good behavior.
And props to the moderators, who apparently think anything with links deserves some points. Bah.
Would there be benefits to managing 20 1048 CPU single-OS systems versus 54,000 linux machines?
Some, but they don't outweigh the costs.
A 2 GB 2 CPU box is well under $1000, right? so 32 of those would be, say, $25,000? But the SGI Altix 3700 configured with just 64 processors and 64 GB of RAM costs a cool million dollars.
Even if the multi-box solution demands more fuss to manage (higher failure rates, I'd guess), since you've saved $975,000, you can afford a little admin time. Plus, the cost of somebody who can build intel boxes must be less than half of somebody who is blessed to work on high-end SGI supercomputers. And I'm sure the SGI service and support contract in't cheap, either.
And the nice thing about a distributed approach like this is that if any given box goes down, so what? Whereas if one of your bazillion-dollar 1048-CPU boxes coughs up a hairball, that's a noticeable percentage of your regional computing capacity.
There are some applications where you can't go for this federated approach, but not as many as people think.
Yep. A single rack, unpowered, at an adequate colo in the bay area is circa $300 per month. Assuming these are 2U boxes, that's about $170 per year in space for a box.
:-)
For dual-processor, multi-disk boxes, the power draw is more than the 60-90 watts; I recently measured a bunch and they ranged from 100-220 watts. (I was impressed that this depends on CPU load; my dual-processor P3 box uses an extra ~10 watts per processor at 100% CPU, even with the disks quiescent.) Anyhow, a 20A circuit at the same colos runs for around $300 / month as well; figure two of those per cabinet and that's another $340 a year per box.
Now I'm sure Google gets the volume discounts, but even at 50% off, it's at least worth it to pay somebody to go turn off broken machines. And even if they just throw out the hardware, pulling should be worth the money, too. So hey, if the Google engineers are reading this, let me just say that I'm glad to come by and pull all of the broken machines out for you at very reasonable rates, and I'll even haul them away for free.
When you offer your trained expert opinion to an idiot, expect nothing less that idiocy in return.
Many technical consultants don't realize that "technical" isn't the only word in their job description. The way I look at it, it's not just my job to give good advice, it's to do so in a way that they understand and follow.
That's not to say your pal wasn't dealing with somebody certifiable. But I'd say only 30-40% of my brainpower when consulting goes into the technical stuff; the rest is spent on understanding and dealing with people, in all their falliable glory.
Back in college I did tech support. We would frequently get calls from people who had monitors with Trinitron tubes saying that two thin black horizontal lines had suddenly appeared after months of ownership.
It took a lot to persuade them that those lines had always been there and that they had just noticed them. Some people would insist on coming in to look at identical monitors just to make sure we weren't lying to them.
That's a good list. The only thing I'd add is to suggest using something like LIDS. The basic notion is that you add something to the kernel so that root is no longer a trusted account. Instead, you give very specific permissions to each program that needs them.
That way even if somebody does compromise, say, named, they won't be able to screw with much. And since LIDS logs attempts to violate security, as soon as they try to, say, view the password file or alter the logs, I'll know about it.
I'll admit that I find spam as annoying as the next guy, but I try to take precautions and use a fairly effective junk mail filter [apple.com], so spam at best rises to the level of minor annoyance. Aren't there better things to spend our money and time lobbying lawmakers and dragging people to court about?
This sounds reasonable. People said it when spam was just getting started. They said it when spam was a few percent of email. And again when it was 10%. Now it's circa 50% and still growing rapidly.
There's no particular reason to believe that this trend will improve. And spam is a classic evolutionary arms race: every advance made in spam defense is met with an advance in spammer offense. And as with any parasite/host relationship, the parasites have a much stronger interest in succeeding (because it's their whole livelihood) than the hosts do (because spam is just one of many problems in their lives).
So our options are A) try to fix it now, or B) wait until it reaches some increased level of annoyance. Personally, I think the billions of dollars wasted on spam each year are enough to justify the effort of extermination.
Oh, great, not only would the moron be breathing down my neck while I was typing, he'd be changing my code to!
Well, if he's changing your code to exclamation points, you need a better partner.
Seriously, pair programming doesn't mean that somebody will be breathing down your neck or changing your code, and the suggestion in the post is stilly.
Real pair programming is a lot like the relationship between a driver and a navigator on a road trip. Back-seat drivers are a pain in the ass, but a good navigator can keep you on the optimal path.
ignoring my revulsion to the idiocy that is Extreme Programming.
Works for me! And a lot of others. I'm sorry to hear it doesn't work for you. Oh, wait, you have tried it before trashing it, right?
And premature optimization has been proven to be very, very bad time and again!
Agreed. And XP is in full agreement with that, too.
Instead of the CEO of McDonalds making 48zillion, he could suck it up, tighten his belt, buy US raised cattle and only make 40zillion. *gasp* a 8 zillion dollar loss. But hey - when you've already got a few billion - what's a zillion?
(yes - I know... zillion... I'm trying to make a point)
Why don't you go out and try this will real numbers? Look up the McDonalds CEO's salary, calculate how many cows they buy, and how much difference him giving up his salary would make?
I'd bet the difference is much smaller than you'd think.
In Japan, a corporate exec's salary is based upon the success (or lack of success) of a company. That is a proper business model. Maybe it would make our business leaders more honest (read: ENRON, etc...).
You should learn a little more about the topic. Simpleminded solutions about making corporate leaders more responsible are one of the major causes of the recent accounting frauds.
Why? Well, it used to be that CEOs were mainly paid via salary. But it was felt that CEOs weren't responsible enough that way, that they tended to do things that were interesting for them, but not necessarily beneficial to customers or shareholders.
So some bright people said, "Gosh, let's make it so that their interested are better aligned with shareholders. We'll give them stock, and stock options. And bonuses, too."
Which in many cases worked. But it also gave a very tempting opportunity to a lot of feckless MBA-fueled locusts, who took the options, bent the corporate numbers to look better than they were, took their fat checks, sold their options, and ran. And you know the story from there.
The lesson? Yes, it's important to make people responsible, especially the executives. But simple, obvious solutions generally come with unintended consequences.
One of the reasons we thrived after WWII was due to the fact that we were a nation that produced goods for the entire world. Every part in those goods were made in the USA, by Americans.
If we made goods for the entire world, what would we get back? The answer has to be "other valuable stuff," otherwise we'll stop doing it. That's trade, and tariffs harm it.
And if we make goods only for ourselves, we end up poorer than we would if we traded. Honest. Go do some research on "comparative advantage", the economic principle that underlies trade. Or imagine what would happen if each state, each city started putting up tariffs to "protect" local industries. A giant clusterfuck, that's what.
It would gratify me immensely if people would go off and actually do a little research before replying. If nothing else, it would be more fun.
If you had, you would have discovered that the median male individual income has increased by about 10%, and the median female income has increased by 90%. I couldn't find median figures together, but US per capita income has gone up 93% since 1968. And all of these numbers are inflation-adjusted, but using CPI, which overstates inflation; actual purchasing power is better still.
And of course, this is with a substantial increase in population. If trade really were destroying jobs, then we would expect there to be huge unemployment, rather than the huge increases in jobs and the standard of living.
As to husbands and wives working, I note that this partly because, gosh, both men and women want to work, and there's much less discrimination against women these days. I know very few stay-at-home moms, and every one of them thinks of it as pretty temporary.
But another important factor is that we have higher expectations. Go look up the number of cars per person in 1968, for example. Or the size of the average home. Or the number of televisions. Most people also feel like they must have about a zillion consumer goods and services, many of which didn't exist or weren't widely available (microwaves, cable tv, computers, mobile phones, personal stereos, car stereos, video games, home gyms, etc, etc, etc).
So for those who are comparing with the golden ages now seen only on Nick at Nite, perhaps if you try living a similar lifestyle, you'll see how much things have improved.
Your stats about the trade increase since the 60s are irrelivant. You forget to add other variables such as inflation.
No, I don't. Take a look at US household income statistics. Adjusted for inflation, the median household income has increased 26% over the same period. And that's using the CPI, which if anything understates things. E.g., consider what kind of computing power (or home entertainment gear, or about eight zillion other things) you could have bought in 1968.
Imagine how much more the dollar would be worth if the clothing companies moved production back to the US, and gave those jobs back to Americans.
If all clothing were made in the US, then there would be a few more US jobs, true at least for a while. But to benefit a very few workers, all US consumers would pay more for their clothing, making the consumer's dollar worth less, not more. And if people are forced to spend more money on clothes, that's less that they spend elsewhere, putting other Americans out of work.
It doesn't take a genius to figure out that if you create jobs, you lower unemployment. if you lower unemployment and create jobs, people have money. If people have money - they spend it. If people spend money - it's good for the economy. *duh*
As H. L. Mencken says, "For every complex problem there is an answer that is clear, simple, and wrong."
It's true that creating jobs is good. But trade barriers don't create jobs; they just move them around. And since they move them from more efficient producers to less efficient ones, they destroy value, making everybody poorer.
Sure. See this paper. Here's part of the abstract:
Although I'm not an economist, my understanding is that the correlation is pretty well established, both theoretically and empirically.
Are they really better or is it because your boss is greedy?
The basic purpose of capitalism is to make greedy people useful to society. Overall, it has worked pretty well. People complain about your namesake, Bill Gates, but I'd rather have him making billions than running mafias or starting land wars, the historically typical pursuit of the ultra-ambitious.
But I don't think people are outsourcing to India because on some absolute scale the Indian programmers are better. I think they do it because they are a better value. To keep our high salaries, we will have to find new ways to earn them.
Part of the reason for the huge slashes is that programmers are viewed as standard commodities that do not bring money into the buttom line. They do not give a shit about your skills if you do not make them money. Therfore an underskilled Indian would be more appropriate then you for the job.
So the interesting question is, are the people who think that correct? If the answer is yes, then programmers need to step up to the plate and start providing business value. And if the answer is no, that programmers do provide plenty of value, then we need to make sure that people understand that.
Also minimum wage laws helped the us economy spreading more wealth back.
There's a lot of disagreement among economists about the effect of minimum wages, but they are generally thought to be net mildly harmful to the economy. If you want to spread wealth, tax policy is a better way to do that.
A much better approach is to find a way to justify the higher wages of American works. In order to not get outsourced, find a way to prove you're worth the money. Is that sheer code productivity, product insights, or what?
My favorite is to make use of short-cycle, agile development methods like Extreme Programming. If you take full advantage of the high-bandwidth, low-latency medium of physical presence, you can achieve things that outsourcing can't match.
generally speaking "free trade" is often just code for "legal loophole"
Sometimes that's true, but it's mainly in the short term.
Take a look at what's happening to formerly low-cost producers like Japan and South Korea, the scary monsters from decades gone by. We all got richer through trade. And rich people don't like to work at crappy jobs, so they are exporting a lot of the low-cost manufacturing they used to do to countries that are still poor. Along the way, South Korea has picked up a lot in the way of democracy and human rights.
totalitarian security
This strikes me as an especially pernicious issue. My first instict is that we shouldn't do business with oppressive regimes. But on the other had, trade with China has been a big moderating influence on them, and the new generation of capitalists has been forcing political reforms that seemed impossible 20 years ago.
I wish I had a better answer than "take a middle path" between economic isolationism and lassez-faire globalization, but it seems that economic wealth and democracy (and all the regard for human rights it entails) are tightly intertwined.
By removing jobs from the US, they're hurting the economy.
Are they?
Trade has increased drastically over the decades, both in absolute terms and as a percentage of GDP (from under 10% in the late 60s to over 25% now). So by your theory, the economy should be a wreck.
But curiously, we're much better off by almost any measure. We're in the middle of a downturn, but unemployment is still under 6%, and the economy's still growing.
So perhaps your theory needs a little revision.