Pretty much the same reason that I haven't shopped in Wal*Mart in about 5 or 6 years now. They're simply too large for the consumer's long-term good. Rather then allowing them to wield more power by adding to their bottom line, I choose to shop elsewhere and encourage others to shop elsewhere as well.
(Plus I know how they treat their suppliers... like chattel to be driven before them.)
When IPV6 is here, Choicepoint will probably pay for your MAC address. And everyone else will pay Choicepoint to know who the "anonymous" person is visiting their website.
At which point, I suspect something will come along that will randomize your MAC address during every reboot. Just like there are privacy tools today that keep cookies off of your system.
Considering that I've recently purchased hand-crafted furniture... hand-crafted wood does not come cheap. It's the not the cost of the wood (unless the maker has to get it from elsewhere), it's the labor and the knowledge of how to put it together.
Don't even plan on talking to a carpenter who makes furniture unless you have a few grand to spare for each piece.
OTOH, you'll end up with a piece of furniture that can easily last 50+ years.
I've used a few of the Asus AM2 motherboards (the microATX M2NPV-VM for our business systems, M2N-E for Linux servers, M2N32-SLI Deluxe for a power server).
All of them have performed well (except for the M2N32-SLI which has issues with Linux using the currently available BIOS from Asus). The 6150 boards (microATX) work well as business desktops and the M2N-E performs very well as a Linux server board (with 6 SATA plugs).
If the M2N-E had dual-gigabit NICs onboard, I'd be estatic about it. We've dropped 4GB of ECC RAM in it and are running Xen with no issues. (Which is the main reason we're putting all AM2 systems in... for the hardware virtualization support.)
Socket F systems have been very slow out of the gate. Only a handful of motherboards and limited availability on the CPU supply side.
We decided to go with dual-core AM2s and ECC RAM for pacifica virtualization support. (Plus we've spent the time to work out the Linux bugaboos on the Asus AM2 motherboards, so we'll stay with the "devil we know" for at least a year.)
I agree, nothing went wrong. The only point that I think that the article was right on was cost. Since the performance difference between the current generation of AM2 processors and 939 processors is so small (or almost negligible), the average consumer is buying based on price. And since 939 processors and systems are still available, though less "desirable" from being "older tech", the prices are usually better on 939.
Now, see, that's where I think the article went wrong. What we've found is that DDR2 memory is slightly cheaper then DDR memory (at least back in July/August) which makes up for the cost of the AM2 motherboard. (In fact, 2GB DDR2 was much cheaper then 2GB of DDR at the time. But memory prices have gone up over the past 4 weeks, so things have changed slightly.)
CPU costs for the Athlon64 X2s are almost identical for the 939 vs AM2 chips. Prices on the 939 3800+ chip are actually up to $175 while the AM2 3800+ chip is selling for below $150 now.
Notes:
- We only buy dual-core AMD chips (ever since the late-July price break)
- I'm comparing Asus motherboards that are pretty much identical except for the memory controller and CPU socket (which keeps things sane because the chipset is already tested and known to work)
If it's 43C under load, you're doing okay. But if it's 43C idle and even higher under load, you're starting to flirt with the 50C+ danger zone. Heck, it's probably better to have a drive that is always 43C then one that goes from 30C to 45C constantly.
I have 2 drives at work (Raptors) that are 30C idle and 32C loaded, but they're in a bay cooler with a 80mm fan that pulls in from the outside and blows directly across the drive. Reckon those drives will last for quite a few years...
Maybe because there hasn't been much news on the drive front this year? The Seagate 750GBs have been out for a while and there's very few new drives out, even in the laptop market (I can think of 2 drives that now use PR tech). Everything else is pretty much the same as it was last year.
SATA connectors seem to have been designed primarily for hot-plug capability in removable drive trays. Some of the newer SATA data cables now come with a metal clip on one side that grips the inside of the SATA data port.
You can generally get 5.25" SATA hotplug backplanes for about $30/disk. Common sizes are 3 trays in (2) 5.25" bays or 4 trays in (3) 5.25" bays. (There's even a 5:3 design but the inside of your 5.25" bays need to be clear of any obstructions along the sides.)
Personally, I like most of the Antec cases (p160, Sonata II, Overture II). I think they're doing a good job of silent / quiet computing by moving things around and putting 120mm fans in.
I think the newer nForce 590 has more PCIe lanes. IIRC, the one motherboard I have with nForce 590 is a 16+16+4+1 for a total of 37 lanes (another note says 46 lanes)
None of the AM2 chips support DDR (the whole reason we needed a new socket was so that you couldn't mix up the DDR2 AMD chips with the DDR AMD chips). Plus all of the AM2 chips (except Sempron) support hardware virtualization (allowing virtual machines using unmodified guest OSs).
Otherwise you can compare the Asus product line and you'll see that the AM2 motherboards look a *lot* like the 939 motherboards. The only difference is hardware virt and the new type of memory.
(I've used both of the 6150 chipset microATX boards, 939 and AM2. We're using them for business workstations. Also own a SLI deluxe AM2 and the M2N-E AM2 motherboard. The M2N-E has better Linux support using the latest official BIOS, for the M2N32-SLI Deluxe you'll need to get ahold of the unofficial 0704 BIOS.)
Dunno, I went with a 7800GS instead as my last upgrade for this AGP gaming system. In another 2 years or so I'll finally replace this system with a 4-core AMD chip and a new PCIe video card.
750GB drives are already using PR. The previous limit was 500GB in a 3.5" form factor. Everything I've read about PR indicates that the limit is probably somewhere around 2.0-2.5TB in a 3.5" drive (only a 4x-5x improvement). Unless they drive that number up a bit more, which might get us to the 3-4TB range.
Ah, that does make a lot more sense then. I forget sometimes how rudimentary things were back in the early 1900s with no frequency allocations or frequency selectivity. I now picture an early version of Ethernet 10baseT with everyone shouting their signal, noting whether it got stepped on, and then retrying their transmission after a random amount of time.
I believe that the FCC was formed to deal with the chaos, but not specifically due to a single incident. That part still seems a little contrived (or correlation vs causation). But then, if the FCC was founded soley based on a single incident, it's just more proof that those who don't learn from history are doomed to repeat it (whipping up public sentiment to create yet another sprawling bureaucracy).
It was said that distress signals from RMS Titanic were stepped-on, garbled, and the transmission of events surrounding the ship's sinking were tough to pass around via radio because there were so many people stepping all over the frequencies. Not sure of the whole story though, so take it as you will.
Sounds like FUD to me. The Titanic was still a long way from civilization (somewhere south of Newfoundland, which is pretty far east of any big cities). Any stepping on would've been done by the other ship operators in the nearby ocean. How chatty were ship's wireless operators back then? How many ships would've been within a few hundred miles?
My assumptions are that 10-year old girls back in 1993 would not likely have access to UNIX since it was almost entirely only available at universities, government offices and large businesses. Most (actually all) 10 year old kids I ever knew did not have sufficient access to such locations as to be able to become familiar with UNIX. And there was very limited net access back then so learning remotely would be difficult too.
AT&T sold a unix SysIV (SysV?) system back in the late 80s with a 20MB hard drive, 5.25" floppy in an attempt to compete with the PC market. Our parents were foolish enough to purchase one (employee discount) and we replaced it the next year with an AT&T 286.
There was all sorts of weird stuff going on in the late 80s as the PC market was still in a state of flux. (Wasn't until sometime in the very early 90s that PCs started becoming known as MS-DOS only.)
The problem is, they have absolutely no justification for any of their numbers. For instance, on page 5 they claim, "In 2008, IDC predicts that 80% of Microsoft client operating systems shipped into enterprises will be Windows Vista." But they can't back it up!
That's probably not a bad guess. Figure that WinV comes out in early 2007 and that most corps will have evaluated it by late 2007. So a 20-80 split between WinXP and WinV sounds about right for enterprise system OSs (that are based on MS). The only holdouts will be the same sorts of corporations that are hanging on to Win2k for the desktop.
(But yes, it's a guess and I don't think you can get much more accurate then +/- 5%.)
Even without the performance advantages, SATA connectors/cables are a heck of a lot more convenient. Modern motherboards have four to six little SATA ports with no master/slave nonsense. The cables/connectors (including power) are so much thinner, easier to work with, and less likely to get loose. Haven't you ever had a boot problem from a bad PATA or power connection to your hard drive (like I have)?
Mmmm, the SATA connector is not the best designed connector in the world in my experience. And that applies to both data and power connectors.
I've never broken a PATA data connector, but I've had issues with SATA connectors being brittle. You have to be very gentle when unplugging SATA data connectors.
Power connectors for SATA are nowhere near as secure as the older 4-pin PATA. Older PATA plugs could require a lot of effort to disconnect. The SATA data connectors can be addressed with the use of the newer locking SATA data cables. But I have yet to see locking connectors for the power plugs. Haven't had a SATA power cable shake loose yet, but it's something I keep in mind.
OTOH... SATA power/data connectors are designed for hotplug capability. Which makes it easy to design and use hot-plug trays with the drives.
(I like the SATA drive features... just wish they'd done it with a better connector design.)
Where I work, it doesn't matter that we can support an OS ourselves, we still need a support contract to CYA.
There are support contracts out there for Gentoo. They're just not as easy to find as the more well-known names like Redhat or SUSE.
Frankly, there's not much support needed for the base O/S. Assuming you can resist the urge to change things willy-nilly or run without backups, almost all of the linux flavors are pretty much the same in terms of stability. Most of the time, you're looking for application level support (Apache, PostgreSQL, Java), not O/S support. And application support contracts are a lot easier to find.
Things that make a system administrator's life easy:
- a written recovery plan
-/etc (and other hand-edited files) placed under version control
- local snapshot-based backup that uses hard-links (quick-n-dirty rollbacks)
- RAID with hot-spares and fast-swap or hot-swap drive bays
- virtualization (moving services to another machine when one goes belly up)
- a more formal backup system (Bacula, etc.)
Notice that none of those are Gentoo-specific. Gentoo just brings portage to the table along with USE flags which is a big benefit.
(Heck, one of the systems that we're building right now is a Xen kernel, Gentoo Dom0, with multiple O/Ss running in DomUs. Which is a bit of a mix-n-match system.)
I've setup close to a dozen systems over the past 2 years with Gentoo (but none of them are desktop units). You're pretty close to the mark. Or at least there's enough truth in that to be funny.
However, because I've done this so many times, I work from a cheat sheet. Which is the web page that I wrote the previous time that documents all of the commands that I used. I still double-check my work every few months against the new handbook, but mostly it's a copy-n-paste between my web browser and my SecureCRT ssh window. That eliminates a lot of typos, and I can keep a log file of the install process if I need to figure out what command I typed wrong.
I haven't quite gotten to the point where I simply clone one setup to another (all of the machines have been specialized to some degree), but I'm close.
Note that the tech improvements slowed down after 2000-2001. Look at specs of a 2002 Tecra 9100 - 1.8GHz, 1GB RAM, 100GB HD, SXGA+ screen. Which is still a very capable machine (but that is slowly showing its age). Next year when it turns 5 I plan on handing it off to a less demanding user who can probably get another 3 years of life out of it.
The days of double-the-performance every 15 months are probably long gone. (Unless AMD and Intel can manage to double the cores in their chips every 15 months.) Nowadays, things run on a longer ~3 year cycle before performance doubles. A good laptop can easily last 5 years if it's taken care of.
So it's more likely to be worth that 10-15% upgrade cost now then it used to be.
I see that someone else also reads the PostgreSQL mailing lists. Of course, the prevailing consensus over there was that it wasn't worth discussing in the [general] list.
Hardware support is easy, software support is a nightmare. Which is why I won't build PCs for regular people. There's no profit in it as soon as you have to start fielding phone calls. Instead I'll point them at Dell / Lenovo / Toshiba / Apple and tell them to get the 3-5 year warranties.
Slashdot Mirror
Pretty much the same reason that I haven't shopped in Wal*Mart in about 5 or 6 years now. They're simply too large for the consumer's long-term good. Rather then allowing them to wield more power by adding to their bottom line, I choose to shop elsewhere and encourage others to shop elsewhere as well.
(Plus I know how they treat their suppliers... like chattel to be driven before them.)
When IPV6 is here, Choicepoint will probably pay for your MAC address. And everyone else will pay Choicepoint to know who the "anonymous" person is visiting their website.
At which point, I suspect something will come along that will randomize your MAC address during every reboot. Just like there are privacy tools today that keep cookies off of your system.
Considering that I've recently purchased hand-crafted furniture... hand-crafted wood does not come cheap. It's the not the cost of the wood (unless the maker has to get it from elsewhere), it's the labor and the knowledge of how to put it together.
Don't even plan on talking to a carpenter who makes furniture unless you have a few grand to spare for each piece.
OTOH, you'll end up with a piece of furniture that can easily last 50+ years.
I've used a few of the Asus AM2 motherboards (the microATX M2NPV-VM for our business systems, M2N-E for Linux servers, M2N32-SLI Deluxe for a power server).
All of them have performed well (except for the M2N32-SLI which has issues with Linux using the currently available BIOS from Asus). The 6150 boards (microATX) work well as business desktops and the M2N-E performs very well as a Linux server board (with 6 SATA plugs).
If the M2N-E had dual-gigabit NICs onboard, I'd be estatic about it. We've dropped 4GB of ECC RAM in it and are running Xen with no issues. (Which is the main reason we're putting all AM2 systems in... for the hardware virtualization support.)
Socket F systems have been very slow out of the gate. Only a handful of motherboards and limited availability on the CPU supply side.
We decided to go with dual-core AM2s and ECC RAM for pacifica virtualization support. (Plus we've spent the time to work out the Linux bugaboos on the Asus AM2 motherboards, so we'll stay with the "devil we know" for at least a year.)
I agree, nothing went wrong. The only point that I think that the article was right on was cost. Since the performance difference between the current generation of AM2 processors and 939 processors is so small (or almost negligible), the average consumer is buying based on price. And since 939 processors and systems are still available, though less "desirable" from being "older tech", the prices are usually better on 939.
Now, see, that's where I think the article went wrong. What we've found is that DDR2 memory is slightly cheaper then DDR memory (at least back in July/August) which makes up for the cost of the AM2 motherboard. (In fact, 2GB DDR2 was much cheaper then 2GB of DDR at the time. But memory prices have gone up over the past 4 weeks, so things have changed slightly.)
CPU costs for the Athlon64 X2s are almost identical for the 939 vs AM2 chips. Prices on the 939 3800+ chip are actually up to $175 while the AM2 3800+ chip is selling for below $150 now.
Notes:
- We only buy dual-core AMD chips (ever since the late-July price break)
- I'm comparing Asus motherboards that are pretty much identical except for the memory controller and CPU socket (which keeps things sane because the chipset is already tested and known to work)
If it's 43C under load, you're doing okay. But if it's 43C idle and even higher under load, you're starting to flirt with the 50C+ danger zone. Heck, it's probably better to have a drive that is always 43C then one that goes from 30C to 45C constantly.
I have 2 drives at work (Raptors) that are 30C idle and 32C loaded, but they're in a bay cooler with a 80mm fan that pulls in from the outside and blows directly across the drive. Reckon those drives will last for quite a few years...
Maybe because there hasn't been much news on the drive front this year? The Seagate 750GBs have been out for a while and there's very few new drives out, even in the laptop market (I can think of 2 drives that now use PR tech). Everything else is pretty much the same as it was last year.
SATA connectors seem to have been designed primarily for hot-plug capability in removable drive trays. Some of the newer SATA data cables now come with a metal clip on one side that grips the inside of the SATA data port.
You can generally get 5.25" SATA hotplug backplanes for about $30/disk. Common sizes are 3 trays in (2) 5.25" bays or 4 trays in (3) 5.25" bays. (There's even a 5:3 design but the inside of your 5.25" bays need to be clear of any obstructions along the sides.)
Define "pretty"?
Personally, I like most of the Antec cases (p160, Sonata II, Overture II). I think they're doing a good job of silent / quiet computing by moving things around and putting 120mm fans in.
I think the newer nForce 590 has more PCIe lanes. IIRC, the one motherboard I have with nForce 590 is a 16+16+4+1 for a total of 37 lanes (another note says 46 lanes)
http://www.sudhian.com/
nForce 550 - 20 lanes
nForce 570 - 20 lanes
nForce 570 SLI - 28 lanes
nForce 590 SLI - 46 lanes
None of the AM2 chips support DDR (the whole reason we needed a new socket was so that you couldn't mix up the DDR2 AMD chips with the DDR AMD chips). Plus all of the AM2 chips (except Sempron) support hardware virtualization (allowing virtual machines using unmodified guest OSs).
Otherwise you can compare the Asus product line and you'll see that the AM2 motherboards look a *lot* like the 939 motherboards. The only difference is hardware virt and the new type of memory.
(I've used both of the 6150 chipset microATX boards, 939 and AM2. We're using them for business workstations. Also own a SLI deluxe AM2 and the M2N-E AM2 motherboard. The M2N-E has better Linux support using the latest official BIOS, for the M2N32-SLI Deluxe you'll need to get ahold of the unofficial 0704 BIOS.)
Memory prices are up over the past 2 weeks which is driving costs up for everything. And the price on the 7800GS isn't *that* bad ($275-$300).
(A dual-core CPU, motherboard, 2GB of RAM is currently running around $500.)
Dunno, I went with a 7800GS instead as my last upgrade for this AGP gaming system. In another 2 years or so I'll finally replace this system with a 4-core AMD chip and a new PCIe video card.
750GB drives are already using PR. The previous limit was 500GB in a 3.5" form factor. Everything I've read about PR indicates that the limit is probably somewhere around 2.0-2.5TB in a 3.5" drive (only a 4x-5x improvement). Unless they drive that number up a bit more, which might get us to the 3-4TB range.
Ah, that does make a lot more sense then. I forget sometimes how rudimentary things were back in the early 1900s with no frequency allocations or frequency selectivity. I now picture an early version of Ethernet 10baseT with everyone shouting their signal, noting whether it got stepped on, and then retrying their transmission after a random amount of time.
I believe that the FCC was formed to deal with the chaos, but not specifically due to a single incident. That part still seems a little contrived (or correlation vs causation). But then, if the FCC was founded soley based on a single incident, it's just more proof that those who don't learn from history are doomed to repeat it (whipping up public sentiment to create yet another sprawling bureaucracy).
It was said that distress signals from RMS Titanic were stepped-on, garbled, and the transmission of events surrounding the ship's sinking were tough to pass around via radio because there were so many people stepping all over the frequencies. Not sure of the whole story though, so take it as you will.
Sounds like FUD to me. The Titanic was still a long way from civilization (somewhere south of Newfoundland, which is pretty far east of any big cities). Any stepping on would've been done by the other ship operators in the nearby ocean. How chatty were ship's wireless operators back then? How many ships would've been within a few hundred miles?
My assumptions are that 10-year old girls back in 1993 would not likely have access to UNIX since it was almost entirely only available at universities, government offices and large businesses. Most (actually all) 10 year old kids I ever knew did not have sufficient access to such locations as to be able to become familiar with UNIX. And there was very limited net access back then so learning remotely would be difficult too.
AT&T sold a unix SysIV (SysV?) system back in the late 80s with a 20MB hard drive, 5.25" floppy in an attempt to compete with the PC market. Our parents were foolish enough to purchase one (employee discount) and we replaced it the next year with an AT&T 286.
There was all sorts of weird stuff going on in the late 80s as the PC market was still in a state of flux. (Wasn't until sometime in the very early 90s that PCs started becoming known as MS-DOS only.)
The problem is, they have absolutely no justification for any of their numbers. For instance, on page 5 they claim, "In 2008, IDC predicts that 80% of Microsoft client operating systems shipped into enterprises will be Windows Vista." But they can't back it up!
That's probably not a bad guess. Figure that WinV comes out in early 2007 and that most corps will have evaluated it by late 2007. So a 20-80 split between WinXP and WinV sounds about right for enterprise system OSs (that are based on MS). The only holdouts will be the same sorts of corporations that are hanging on to Win2k for the desktop.
(But yes, it's a guess and I don't think you can get much more accurate then +/- 5%.)
Even without the performance advantages, SATA connectors/cables are a heck of a lot more convenient. Modern motherboards have four to six little SATA ports with no master/slave nonsense. The cables/connectors (including power) are so much thinner, easier to work with, and less likely to get loose. Haven't you ever had a boot problem from a bad PATA or power connection to your hard drive (like I have)?
Mmmm, the SATA connector is not the best designed connector in the world in my experience. And that applies to both data and power connectors.
I've never broken a PATA data connector, but I've had issues with SATA connectors being brittle. You have to be very gentle when unplugging SATA data connectors.
Power connectors for SATA are nowhere near as secure as the older 4-pin PATA. Older PATA plugs could require a lot of effort to disconnect. The SATA data connectors can be addressed with the use of the newer locking SATA data cables. But I have yet to see locking connectors for the power plugs. Haven't had a SATA power cable shake loose yet, but it's something I keep in mind.
OTOH... SATA power/data connectors are designed for hotplug capability. Which makes it easy to design and use hot-plug trays with the drives.
(I like the SATA drive features... just wish they'd done it with a better connector design.)
Where I work, it doesn't matter that we can support an OS ourselves, we still need a support contract to CYA.
/etc (and other hand-edited files) placed under version control
There are support contracts out there for Gentoo. They're just not as easy to find as the more well-known names like Redhat or SUSE.
Frankly, there's not much support needed for the base O/S. Assuming you can resist the urge to change things willy-nilly or run without backups, almost all of the linux flavors are pretty much the same in terms of stability. Most of the time, you're looking for application level support (Apache, PostgreSQL, Java), not O/S support. And application support contracts are a lot easier to find.
Things that make a system administrator's life easy:
- a written recovery plan
-
- local snapshot-based backup that uses hard-links (quick-n-dirty rollbacks)
- RAID with hot-spares and fast-swap or hot-swap drive bays
- virtualization (moving services to another machine when one goes belly up)
- a more formal backup system (Bacula, etc.)
Notice that none of those are Gentoo-specific. Gentoo just brings portage to the table along with USE flags which is a big benefit.
(Heck, one of the systems that we're building right now is a Xen kernel, Gentoo Dom0, with multiple O/Ss running in DomUs. Which is a bit of a mix-n-match system.)
that's the first one
LOL
I've setup close to a dozen systems over the past 2 years with Gentoo (but none of them are desktop units). You're pretty close to the mark. Or at least there's enough truth in that to be funny.
However, because I've done this so many times, I work from a cheat sheet. Which is the web page that I wrote the previous time that documents all of the commands that I used. I still double-check my work every few months against the new handbook, but mostly it's a copy-n-paste between my web browser and my SecureCRT ssh window. That eliminates a lot of typos, and I can keep a log file of the install process if I need to figure out what command I typed wrong.
I haven't quite gotten to the point where I simply clone one setup to another (all of the machines have been specialized to some degree), but I'm close.
Note that the tech improvements slowed down after 2000-2001. Look at specs of a 2002 Tecra 9100 - 1.8GHz, 1GB RAM, 100GB HD, SXGA+ screen. Which is still a very capable machine (but that is slowly showing its age). Next year when it turns 5 I plan on handing it off to a less demanding user who can probably get another 3 years of life out of it.
The days of double-the-performance every 15 months are probably long gone. (Unless AMD and Intel can manage to double the cores in their chips every 15 months.) Nowadays, things run on a longer ~3 year cycle before performance doubles. A good laptop can easily last 5 years if it's taken care of.
So it's more likely to be worth that 10-15% upgrade cost now then it used to be.
I love my Yugo luge commute
I see that someone else also reads the PostgreSQL mailing lists. Of course, the prevailing consensus over there was that it wasn't worth discussing in the [general] list.
Hardware support is easy, software support is a nightmare. Which is why I won't build PCs for regular people. There's no profit in it as soon as you have to start fielding phone calls. Instead I'll point them at Dell / Lenovo / Toshiba / Apple and tell them to get the 3-5 year warranties.