Except after the tennager's year of experimentation, if the kids decide they like the boom boxes, they have to leave their homes and are shunned forever by their families.
People think the Amish are all sweet and nice because they make sturdy furniture, but there's a fanatical side too. Taliban-style sexism, severe corporal punishment of children, racism, etc. are part of the Amish lifestyle, too.
How about a legally binding contract. That's what the license agreement is. Is that worth "much"?
The license is free (as in cost), and only revocable by either party under very specific circumstances (like you selling software contraining the PRA alogrithm without getting a commercial license).
IF MS were to attempt to charge for the use of the PRA algorithm after you agree to the initial license, you could sue them for breach.
It's all detailed there. RTFA, RTFL, and take off your tinfoil hat...
Not to mention that, at least in the U.S., mostly lower-income housing is located next to antenna facilites and power lines. Newer, "richer" neighborhoods typically have no unsightly towers and power lines are all buried.
Unfortunately, these same lower-income folks are the prime target market for fast food chains, beer and liquor companies, tobacco companies, corner drug dealers, etc.
So who's to say it's not the less-healthy lifestyle of those living near antenna facilites and power lines that's causing the statistical bump in cancer rates?
It seems to me that fusion research in the US is never going to get decent levels of funding all the time that the Whitehouse is full of people with millions of dollars invested in oil companies.
Uh, dude... everyone invests in oil companies. I'll bet a not-insignificant piece of your pension/portfolio is in the energy sector. You may not know, because it's wrapped in a pension or mutual fund, but you're almost certainly an "oil company profiteer" to a certain degree.
BP, Exxon-Mobil, etc. are spending billions working on alternative energy. They're not stuipd. You think they want to do business in places like Persian Gulf? The overhead costs of transport and security and the overall business risks are sky-high. But oil is still cheaper than solar/wind/whatever at current technology levels. Energy companies will gladly switch gears to providing us solar-produced hydrogen cells when it becomes economically feasible.
I'd also like to point out that games are written for platforms, not platforms for games. The G5 is an excellant gaming platform with not as many games written for it.
How can an "excellent gaming platform" have only one mouse button?!?
I never understood why "Jap" or "Nip" would be offensive to a Japanese person. They're simply abbreviations of "Japan" and "Nippon", the Anglicized and native names for the country.
I mean, I'm not offended by the hordes of Europeans that call Americans "Amis".
I'll grant you that the terms "Jap" and "Nip" had a decidedly negative connotation amongst Americans during and immediately after WWII, but not because the terms themselves were offensive or deragatory in nature. The terms had a negative connotation commensurate with "Japan","Nippon", and "Germany" at the time - simply because the words represented the enemy in a brutal and costly war.
Now, I agree that other Asian peoples should be offended by being called "Jap" or "Nip", as it is inaccurate. It would be the same as a native in Singapore referring to all persons of European descent as "Brits".
I sitll purchase Intel for my company because I got burned with a bunch of first
K6s, then Athalons that melted when their fans stopped.
As far as I know, AMD has still not introduced clock speed scaling to prevent overheats. Until they do, I will be buying Intel desktops and servers. Allowing the failure of a $1 part to ruin a $300+ CPU is stupid. And alarms don't cut it... lusers just sit there thinking "what's that noise?"
We have run Exchange since version 5.0 on many servers, and have never had an information store become corrupted. It simply does not happen "frequently", at least on decent hardware. Exchange 2000 and 2003 simply have never gone down on us, ever.
Contacts are NOT stored separately from any of the other private mailbox storage in the Exchange system. Public folders are separated from mailboxes. Exchange 2000 introduced the "stream store" for storing messages from each type of store in native RFC2822 format, but everything appears logically the same to the user (and the backup software). And in Exchange 2000/2003, you can have multiple independent stores. Taking one down does not take down the others in the Storage Group. Do you really know anything about Exchange, or do you just sell your services against it?
Not sure why you consider integration "bad design", especially since those functions are all necessary for a business communications tool. You need your contacts handy to do messaging, and you need your calendar handy too. Clicking all over the place and logging into different apps to acomplish this is stupid. There are plenty of 3rd party applications that integrate with exchange. The API and object model suck, I'll grant you that, but they're publicly documented and certainly no more convoluted than those provided by Notes.
Yes, surprise, Exchange costs money. Quite a bit for large multi-server organizations. All those evil commmercial software vendors price software this way. But since Exchange has no true competeition in the OSS community, it will probably continue to do well.
There are plenty of huge, multi-national Exchange enterprises out there. Some have hundreds of thousands of users, and 5000 or more per server. They're not all having the same trouble with the product you claim to have experienced. Maybe you just don't know as much about Exchange as you think you do.
and a database-driven message store (which unlike Exchange, can be backed up hot).
You have absolutely, positively no freaking clue what you are talking about.
From its very first release, Exchange server has had a fully transaction-logged database capable of online backups, and point-in-time recovery. Since version 5.5 (1998?), it has been possible to do clustering with failover hot spare servers.
Do you know anything about this Exchange Server of which we speak?
You're missing something. Bagdad may have had no effective fighter cover, but you don't need fighters for effective air defense.
Badghdad had arguably the heaviest surface-to-air defenses in the history of the world before the air war began in 1991. I read this in CNN's history of the gulf war. I recall the book reporting Baghdad had more anti-aircraft batteries (SAMs and Triple-A) per square km than Hanoi during the Vietnam confict, or Berlin during WWII. And these defensive weapons were far more accurate and lethal than 1960s or 1940s technology. I wish I still had the book to give a direct quote.
The benefit of stealth technology is in defense from SAMs and Triple-A, not in defense from fighters. In fact, an F-117 would proabably be dead meat if it ever got mixed up in a dogfight with a conventional fighter less than 30 years old. F-117s are delicate, have poor maneuverability, and require quite a bit of computer correction on the controls just to remain in stable flight. That's why they fly mostly at night and are painted black.
Basically, stealth technology worked as designed and advertised in 1991, and again in 2003.
I don't really see the need to put things like 'select * from story where storynum=956' in a stored procedure. Looking through most of my code, I would guess that at least 90% of my queries are similar to that.
If you're actually using "SELECT *" in an application's code, you should be fired.
It was a real kick in the pants to shell out $5000 to $15,000 on Xeon systems that had a slower FSB than the $500 desktop counterparts!
You're forgetting something... that there are four 533 MHz memory buses on the big 4-way Xeon chipset. 533 Mhz quad-channel beats the pants of single-or-dual-channel 800 MHz buses any day.
We bought a 4x3.06 GHz system a few months back. Each CPU as 4 MB of cache, and we have 12 GB of RAM in it. It cost as much as a new BMW 325i ($4000+ per CPU!), but it definitely blows away the 2x3.06 GHz boxes we have when running Sandra and other synthetic benchmarks. It's also a helluva lot faster running the database application we bought it for.
Intel does know what it's doing with the Xeon. They didn't kill off most of the RISC/Unix market by screwing their server customers with poor performing and overpriced machines. Indeed, even the $4K top-o-the-line Xeon is an absolute bargain in terms of price/performance when compared to the IBM Power series or what is left of the other competition. That said, the Opteron has changed things a bit. As soon as a version of Windows is available for Opteron that supports >4GB RAM (as the Xeon does with PAE), Intel is going to have to drastically lower Xeon prices. Right now, they still have the only non-Itanium solution that can run Windows server with >4GB of memory.
And yes, you can say "run linux on the Opteron instead", but the fact is a huge number of application vendors require MS SQL server, because they don't want the costs of supporting more than one DB. If you want that vendor's application, and more importantly their support, you have to run Windows servers. If you have an application that can run on DB2 or Oracle on top of Linux, then the Opteron makes a lot of sense.
It would take a constant ready force of 120000+ troops to guard the US borders with Mexico and Canada, with troops spaced 100 M apart.
Multiply that times 3 because troops can only guard 80 hours or so a week to be effective. And multiply by 3 again to include officers, support & logistics personnel.
So it would take a constant standing force of nearly 1 million to "close the US borders." That's a huge chunk of our military, and a huge chunk of change per day. Completely impractical.
BitTorrent has so much setup overhead that's it's silly to use it for small files like RSS feeds. You have to connect to a tracker, get a list of peers, and wait for a peer to optimistically unchoke you. Just the "connect to tracker" part of the BitTorrent handshake probable requires as much work for a server as just sending out the RSS over HTTP. So you would be trading a slashdotting of your web server for a slashdotting of your BitTorrent tracker.
Also, using BitTorrent for RSS doesn't solve the firewall problem, which is why other "push" approaches to RSS distribution won't work. Most enterprises are not going to allow any type of push protocol into their networks, and 95% of home users won't be able to figure out how to do all the firewall shenanigans necessary to make BitTorrent work.
It seems to me that everybody on Slashdot wants to use BitTorrent for everything these days, even though BitTorrent is only good at one thing: decreasing the bandwidth required for distributing large files (not small ones).
I think you're getting your kilobits-per-second and kilobytes-per-second mixed up, my friend.
5-6 kilobytes per second (which is what IE and BitTorrent report) is very near 50 kilobits per second (which is how your connection is rated) when protocol overhead is added.
I just got about 310 kilobytes per second on my 3 megabits-per-second connection when downloading Fedora Core 2 distribution DVD. That's a little over 80% of the rated speed of my wire, about as good as it gets with smallish packets and protocol overhead.
Do one thing, do it well- and never have to worry about pleasing everyone, having conflicting goals, etc.
A smart business/organization/whatever should never follow your advice. They should be at least somewhat diversified, and do at least a few things very well.
There's a reason Polaroid is a shell of its former self: Polaroid wasn't diversified at all. They did just one thing very well (instant cameras and supplies), and when technology and the market changed rapidly, Polaroid became irrelevant.
I'm not sure how this principle applies to Debian or Linux distros in general, but if Debian (the organization) wants to survive long-term, it will ultimately need to fulfill more than one market niche.
Bayesian filtering is dead now with the random garbage-spewers, so I need to test and install another solution on the server end (until the last 6 months or so, client filtering worked best for me - now it sucks ass).
I use SpamBayes, and I got 80 spam yesterday. 78 were filtered automatically into my spam folder, and 2 were sent to my "unsure" folder for manual attention. I had no false positives.
Bayesian filtering does work, even with those "random garbage spewers". But the cost of receiving, storing, and filtering those spam were still borne by my company. Those spam were 276 KB, so let's double that and call it 552 KB of network traffic. That's 0.0000175% of hte daily capacity of my company's 3 Mbps pipe. We pay US $60/day for that connection, so receiving that spam cost me just US$0.001. We have 100 users, and I know they get less spam than me, but let's call it $0.10 per day for spam bandwidth costs for my whole company. $3 a month.
Now storage costs are also minimal, our SAN arrays are about $2.50 per gigabyte. So we have 276 KB * 100 users * 30 days = $1.90 to store a month's worth of spam for my company.
CPU power spent on filtering is too cheap to worry about.
Employee time is the real cost. Say each employee has to sort 5 spams per day after Bayesian filtering, at an average of 5 seconds each, and an average hourly cost of $50 to the company. That's $1040 per month in employee time spent dealing with spam, even with the best filtering available. Add in the hours I spend chasing down phishing attacks and administering the spam filters, and we're talking a few grand per month spent on spam by my smallish company!
I manage 25 windows servers, and have not seen an OS-level crash since 2000. Nor have I seen a Microsoft server application crash (we run a lot of SQL Server and Exchange boxes).
Even our IIS applications don't crash frequently, maybe twice a year apiece. And when they do, it's almost always directly traceable to some unstable thrid-party COM object with a memory leak that hasn't been fixed yet. You can set IIS to re-start itself when this happens; this is no different from having a runaway apache module downing apache.
I'd have to say you've got some pretty screwed up hardware and drivers if you're seeing Windows 2000 crash "ALL THE TIME".
Patch deployment is a challenge, but we test everything for a few weeks on test systems before installation, and use MS Software Update Services to deploy the patches hands-free. We haven't had a patch screw up a system or application yet.
I think you're blaming MS for your own poor systems management and your own poor test methodology.
Here's a PSINFO for one of my SQL boxes, which doesn't need IE patches (nobody browses from it) and so hasn't needed to be rebooted since the ASN.1 patch, IIRC:
System information for \\acc-db-chi: Uptime: 63 days 2 hours 54 minutes 31 seconds Kernel version: Microsoft Windows 2000, Multiprocessor Free Product type: Server Product version: 5.0 Service pack: 4 Kernel build number: 2195 Install date: 1/8/2001, 10:02:44 PM
Because, in the AMD formula above, the "work per clock cycle" is the same for each manufacturer.
No it absolutely, positively is not. Any AMD Athalon chip executes more instructions per clock cycle than a Pentium 4. A Pentium M executes more instructions per cycle than a Pentium 4. This is why an AMD chip can be (in the case of Opteron, significantly) faster than an Intel P4 running real programs while limping along at 60% of the P4's clock speed.
I think you need some education on basic computer architecture, my firend. If you read this, you'll understand why a massively super-scalar ("wide") CPU like the Opteron is faster than a deeply pipelined CPU like the P4 on a clock-per-clock basis.
So if an AMD chip running at 2.0 GHz can perform say ~2.4 floating point additions per clock cycle on average, it will be faster (for an FP-ADD heavy application) than a 3.0 GHz P4 which only performs ~1.2 floating point additions per clock cycle.
...correct me if I'm wrong, but I've never seen a hot-swappable RAID 1+0 solution...
You're wrong. Hot-swap ability is a function of the RAID controller and the drive's mounting hardware (or 'cage'). It has nothing to do with the RAID level at all (except that a RAID 0 array can't be hot-swapped, since it has no redundancy and removing a drive would take down the whole array).
We have a bunch of RAID 0+1 systems, some use 80-pin SCSI hot-swap disks and cages, some use ATA hot-swap cages. A wide variety of manufacturers make hot-swap cages for 80-pin SCSI (SCA), Ultra-ATA, and Serial-ATA drives (DataStor, Adaptec, Promise, and SuperMicro to name a few).
And of course the bigger server manufactures make their own with hot-swap cages as well to build into their servers.
Such a system would still be vulnerable to a dictionary attack, which a spammer that controls several thousand zombied PCs could handle quickly and with a fair amount of anonimity for the spammer.
They'll try 'aaa@yahoo.com', 'aab@yahoo.com', 'aac@yahoo.com' until they get hits of addresses on the do-not-spam list. Spammers already do this qutie a bit, and they're smatter about it, using common family and given names (presumably taken from the phone book). I spam-trap attempted mailings to addresses like 'jsmith@mydomain.com' all the time, even though there has never been a jsmith at my domain.
We hired a beige-box manufacturer. We specced it out to various places, and PC Mall built them for the best price. If I had to do it over again, I'm not sure that I wouldn't go with IBM -- while they cost a lot more, I expect that they'd build more solid systems.
Did you talk to PC Mall salespeople for this service? I can't seem to find a beige-box configurator on their site.
A few years ago, I got pretty sick of overpaying for Dell and HP hardware that had cheap disks and RAM that would go belly up after 90 days. We've been building our own servers with chassis and motherboards from SuperMicro, and have had good results in terms of reliability and especially cost. But we would like to get someone to do the assembly grunt work for us if we could spec out all the RAID controllers, disks, etc. ourselves.
What other beige box manufacturers did you look at, besides PC Mall?
Imagine a world in which paper or metal money doesn't exist anymore..would you trust banks/govts/corporations to have all your money in their hands, stored as numbers in their computers ?
Hold onto your tinfoil hat, buddy - those big, bad banks and governments have already gone and done it!
Only a very small percentage of the money floating aroung the world economy is held as actual paper currency.
Do you think your bank has a $1 note sitting in a vault somewhere for every dollar in your account?
We're not on the gold standard anymore. Read up on the concept of fiat money, it is the idea that money has value because of the credit-worthiness of the organization issuing it.
I can say "I have $1000" even though it is only because Citibank has a string of numbers representing $1000 on my behalf. I know Citibank is trustworthy enough to cough up the dough, even converting it into paper money if I wish. Expand this idea to the whole international, interconnected web of individuals, governments, business, and currencies, and that's what money is: trust. A belief that something has value, because other people will give you things for it.
Money is not currency, and the overwhelming majority of money exists in non-currency forms such as accounts receivable or other "stored numbers in computers."
Slashdot Mirror
Except after the tennager's year of experimentation, if the kids decide they like the boom boxes, they have to leave their homes and are shunned forever by their families.
People think the Amish are all sweet and nice because they make sturdy furniture, but there's a fanatical side too. Taliban-style sexism, severe corporal punishment of children, racism, etc. are part of the Amish lifestyle, too.
How about a legally binding contract. That's what the license agreement is. Is that worth "much"?
The license is free (as in cost), and only revocable by either party under very specific circumstances (like you selling software contraining the PRA alogrithm without getting a commercial license).
IF MS were to attempt to charge for the use of the PRA algorithm after you agree to the initial license, you could sue them for breach.
It's all detailed there. RTFA, RTFL, and take off your tinfoil hat...
Not to mention that, at least in the U.S., mostly lower-income housing is located next to antenna facilites and power lines. Newer, "richer" neighborhoods typically have no unsightly towers and power lines are all buried.
Unfortunately, these same lower-income folks are the prime target market for fast food chains, beer and liquor companies, tobacco companies, corner drug dealers, etc.
So who's to say it's not the less-healthy lifestyle of those living near antenna facilites and power lines that's causing the statistical bump in cancer rates?
BP, Exxon-Mobil, etc. are spending billions working on alternative energy. They're not stuipd. You think they want to do business in places like Persian Gulf? The overhead costs of transport and security and the overall business risks are sky-high. But oil is still cheaper than solar/wind/whatever at current technology levels. Energy companies will gladly switch gears to providing us solar-produced hydrogen cells when it becomes economically feasible.
How can an "excellent gaming platform" have only one mouse button?!?
I never understood why "Jap" or "Nip" would be offensive to a Japanese person. They're simply abbreviations of "Japan" and "Nippon", the Anglicized and native names for the country.
I mean, I'm not offended by the hordes of Europeans that call Americans "Amis".
I'll grant you that the terms "Jap" and "Nip" had a decidedly negative connotation amongst Americans during and immediately after WWII, but not because the terms themselves were offensive or deragatory in nature. The terms had a negative connotation commensurate with "Japan","Nippon", and "Germany" at the time - simply because the words represented the enemy in a brutal and costly war.
Now, I agree that other Asian peoples should be offended by being called "Jap" or "Nip", as it is inaccurate. It would be the same as a native in Singapore referring to all persons of European descent as "Brits".
I sitll purchase Intel for my company because I got burned with a bunch of first K6s, then Athalons that melted when their fans stopped.
As far as I know, AMD has still not introduced clock speed scaling to prevent overheats. Until they do, I will be buying Intel desktops and servers. Allowing the failure of a $1 part to ruin a $300+ CPU is stupid. And alarms don't cut it... lusers just sit there thinking "what's that noise?"
Your experiences do not match mine.
There are plenty of huge, multi-national Exchange enterprises out there. Some have hundreds of thousands of users, and 5000 or more per server. They're not all having the same trouble with the product you claim to have experienced. Maybe you just don't know as much about Exchange as you think you do.
You have absolutely, positively no freaking clue what you are talking about.
From its very first release, Exchange server has had a fully transaction-logged database capable of online backups, and point-in-time recovery. Since version 5.5 (1998?), it has been possible to do clustering with failover hot spare servers.
Do you know anything about this Exchange Server of which we speak?
You're missing something. Bagdad may have had no effective fighter cover, but you don't need fighters for effective air defense.
Badghdad had arguably the heaviest surface-to-air defenses in the history of the world before the air war began in 1991. I read this in CNN's history of the gulf war. I recall the book reporting Baghdad had more anti-aircraft batteries (SAMs and Triple-A) per square km than Hanoi during the Vietnam confict, or Berlin during WWII. And these defensive weapons were far more accurate and lethal than 1960s or 1940s technology. I wish I still had the book to give a direct quote.
The benefit of stealth technology is in defense from SAMs and Triple-A, not in defense from fighters. In fact, an F-117 would proabably be dead meat if it ever got mixed up in a dogfight with a conventional fighter less than 30 years old. F-117s are delicate, have poor maneuverability, and require quite a bit of computer correction on the controls just to remain in stable flight. That's why they fly mostly at night and are painted black.
Basically, stealth technology worked as designed and advertised in 1991, and again in 2003.
If you're actually using "SELECT *" in an application's code, you should be fired.
Dude, you might have had a point, but then you went and used "Saturn" and "rims" in the same sentence.
Put down that DVD of 2 Fast 2 Furious and slowly back away...
You're forgetting something... that there are four 533 MHz memory buses on the big 4-way Xeon chipset. 533 Mhz quad-channel beats the pants of single-or-dual-channel 800 MHz buses any day.
We bought a 4x3.06 GHz system a few months back. Each CPU as 4 MB of cache, and we have 12 GB of RAM in it. It cost as much as a new BMW 325i ($4000+ per CPU!), but it definitely blows away the 2x3.06 GHz boxes we have when running Sandra and other synthetic benchmarks. It's also a helluva lot faster running the database application we bought it for.
Intel does know what it's doing with the Xeon. They didn't kill off most of the RISC/Unix market by screwing their server customers with poor performing and overpriced machines. Indeed, even the $4K top-o-the-line Xeon is an absolute bargain in terms of price/performance when compared to the IBM Power series or what is left of the other competition. That said, the Opteron has changed things a bit. As soon as a version of Windows is available for Opteron that supports >4GB RAM (as the Xeon does with PAE), Intel is going to have to drastically lower Xeon prices. Right now, they still have the only non-Itanium solution that can run Windows server with >4GB of memory.
And yes, you can say "run linux on the Opteron instead", but the fact is a huge number of application vendors require MS SQL server, because they don't want the costs of supporting more than one DB. If you want that vendor's application, and more importantly their support, you have to run Windows servers. If you have an application that can run on DB2 or Oracle on top of Linux, then the Opteron makes a lot of sense.
It would take a constant ready force of 120000+ troops to guard the US borders with Mexico and Canada, with troops spaced 100 M apart.
Multiply that times 3 because troops can only guard 80 hours or so a week to be effective. And multiply by 3 again to include officers, support & logistics personnel.
So it would take a constant standing force of nearly 1 million to "close the US borders." That's a huge chunk of our military, and a huge chunk of change per day. Completely impractical.
BitTorrent has so much setup overhead that's it's silly to use it for small files like RSS feeds. You have to connect to a tracker, get a list of peers, and wait for a peer to optimistically unchoke you. Just the "connect to tracker" part of the BitTorrent handshake probable requires as much work for a server as just sending out the RSS over HTTP. So you would be trading a slashdotting of your web server for a slashdotting of your BitTorrent tracker.
Also, using BitTorrent for RSS doesn't solve the firewall problem, which is why other "push" approaches to RSS distribution won't work. Most enterprises are not going to allow any type of push protocol into their networks, and 95% of home users won't be able to figure out how to do all the firewall shenanigans necessary to make BitTorrent work.
It seems to me that everybody on Slashdot wants to use BitTorrent for everything these days, even though BitTorrent is only good at one thing: decreasing the bandwidth required for distributing large files (not small ones).
I think you're getting your kilobits-per-second and kilobytes-per-second mixed up, my friend.
5-6 kilobytes per second (which is what IE and BitTorrent report) is very near 50 kilobits per second (which is how your connection is rated) when protocol overhead is added.
I just got about 310 kilobytes per second on my 3 megabits-per-second connection when downloading Fedora Core 2 distribution DVD. That's a little over 80% of the rated speed of my wire, about as good as it gets with smallish packets and protocol overhead.
A smart business/organization/whatever should never follow your advice. They should be at least somewhat diversified, and do at least a few things very well.
There's a reason Polaroid is a shell of its former self: Polaroid wasn't diversified at all. They did just one thing very well (instant cameras and supplies), and when technology and the market changed rapidly, Polaroid became irrelevant.
I'm not sure how this principle applies to Debian or Linux distros in general, but if Debian (the organization) wants to survive long-term, it will ultimately need to fulfill more than one market niche.
How about this amatuerish effort.
I use SpamBayes, and I got 80 spam yesterday. 78 were filtered automatically into my spam folder, and 2 were sent to my "unsure" folder for manual attention. I had no false positives.
Bayesian filtering does work, even with those "random garbage spewers". But the cost of receiving, storing, and filtering those spam were still borne by my company. Those spam were 276 KB, so let's double that and call it 552 KB of network traffic. That's 0.0000175% of hte daily capacity of my company's 3 Mbps pipe. We pay US $60/day for that connection, so receiving that spam cost me just US$0.001. We have 100 users, and I know they get less spam than me, but let's call it $0.10 per day for spam bandwidth costs for my whole company. $3 a month.
Now storage costs are also minimal, our SAN arrays are about $2.50 per gigabyte. So we have 276 KB * 100 users * 30 days = $1.90 to store a month's worth of spam for my company.
CPU power spent on filtering is too cheap to worry about.
Employee time is the real cost. Say each employee has to sort 5 spams per day after Bayesian filtering, at an average of 5 seconds each, and an average hourly cost of $50 to the company. That's $1040 per month in employee time spent dealing with spam, even with the best filtering available. Add in the hours I spend chasing down phishing attacks and administering the spam filters, and we're talking a few grand per month spent on spam by my smallish company!
I manage 25 windows servers, and have not seen an OS-level crash since 2000. Nor have I seen a Microsoft server application crash (we run a lot of SQL Server and Exchange boxes).
Even our IIS applications don't crash frequently, maybe twice a year apiece. And when they do, it's almost always directly traceable to some unstable thrid-party COM object with a memory leak that hasn't been fixed yet. You can set IIS to re-start itself when this happens; this is no different from having a runaway apache module downing apache.
I'd have to say you've got some pretty screwed up hardware and drivers if you're seeing Windows 2000 crash "ALL THE TIME".
Patch deployment is a challenge, but we test everything for a few weeks on test systems before installation, and use MS Software Update Services to deploy the patches hands-free. We haven't had a patch screw up a system or application yet.
I think you're blaming MS for your own poor systems management and your own poor test methodology.
Here's a PSINFO for one of my SQL boxes, which doesn't need IE patches (nobody browses from it) and so hasn't needed to be rebooted since the ASN.1 patch, IIRC:
No it absolutely, positively is not. Any AMD Athalon chip executes more instructions per clock cycle than a Pentium 4. A Pentium M executes more instructions per cycle than a Pentium 4. This is why an AMD chip can be (in the case of Opteron, significantly) faster than an Intel P4 running real programs while limping along at 60% of the P4's clock speed.
I think you need some education on basic computer architecture, my firend. If you read this, you'll understand why a massively super-scalar ("wide") CPU like the Opteron is faster than a deeply pipelined CPU like the P4 on a clock-per-clock basis.
So if an AMD chip running at 2.0 GHz can perform say ~2.4 floating point additions per clock cycle on average, it will be faster (for an FP-ADD heavy application) than a 3.0 GHz P4 which only performs ~1.2 floating point additions per clock cycle.
You're wrong. Hot-swap ability is a function of the RAID controller and the drive's mounting hardware (or 'cage'). It has nothing to do with the RAID level at all (except that a RAID 0 array can't be hot-swapped, since it has no redundancy and removing a drive would take down the whole array).
We have a bunch of RAID 0+1 systems, some use 80-pin SCSI hot-swap disks and cages, some use ATA hot-swap cages. A wide variety of manufacturers make hot-swap cages for 80-pin SCSI (SCA), Ultra-ATA, and Serial-ATA drives (DataStor, Adaptec, Promise, and SuperMicro to name a few). And of course the bigger server manufactures make their own with hot-swap cages as well to build into their servers.Such a system would still be vulnerable to a dictionary attack, which a spammer that controls several thousand zombied PCs could handle quickly and with a fair amount of anonimity for the spammer.
They'll try 'aaa@yahoo.com', 'aab@yahoo.com', 'aac@yahoo.com' until they get hits of addresses on the do-not-spam list. Spammers already do this qutie a bit, and they're smatter about it, using common family and given names (presumably taken from the phone book). I spam-trap attempted mailings to addresses like 'jsmith@mydomain.com' all the time, even though there has never been a jsmith at my domain.
Did you talk to PC Mall salespeople for this service? I can't seem to find a beige-box configurator on their site.
A few years ago, I got pretty sick of overpaying for Dell and HP hardware that had cheap disks and RAM that would go belly up after 90 days. We've been building our own servers with chassis and motherboards from SuperMicro, and have had good results in terms of reliability and especially cost. But we would like to get someone to do the assembly grunt work for us if we could spec out all the RAID controllers, disks, etc. ourselves.
What other beige box manufacturers did you look at, besides PC Mall?
Hold onto your tinfoil hat, buddy - those big, bad banks and governments have already gone and done it!
Only a very small percentage of the money floating aroung the world economy is held as actual paper currency.
Do you think your bank has a $1 note sitting in a vault somewhere for every dollar in your account?
We're not on the gold standard anymore. Read up on the concept of fiat money, it is the idea that money has value because of the credit-worthiness of the organization issuing it.
I can say "I have $1000" even though it is only because Citibank has a string of numbers representing $1000 on my behalf. I know Citibank is trustworthy enough to cough up the dough, even converting it into paper money if I wish. Expand this idea to the whole international, interconnected web of individuals, governments, business, and currencies, and that's what money is: trust. A belief that something has value, because other people will give you things for it.
Money is not currency, and the overwhelming majority of money exists in non-currency forms such as accounts receivable or other "stored numbers in computers."