Yeah but those poles can generally hold the weight of another cable no problem and if you offer the right incentives to a municipality they will generally agree to enter into a franchise agreement with you. I know I personally use an overlay provider and love not having to deal with a monopoly provider (they actually refused to hook up my house because I had just moved in and didn't know where the grounding rod was for my electrical system!).
Yes but IAPV is also very common in non-collapsed hives so it is believed that it has to be a combination of factors with IAPV being but one (possibly major) factor.
You can definitly get the SAS drives in 15K, I use them for my Citrix blade servers (HP BL460C). During heavy login periods the 15K is a must, especially since the blade only allows two spindles.
They have 20% spare capacity which IS a lot more than you would put in a user servicable enclosure. The great thing is you get the speed advantage of the spares because while they limit the amount of data use to 80% they are normally using all spindles in the pack.
Can your homebrew solution scale from 1TB to 1.something Exabytes with 64GB of cache all without any downtime? Oh and does it come with a 5 year warranty? Also what do you do when your box is at a remote datacenter on another continent and you can't get ahold of an operator?
The broader issue is that primary and secondary education vary enormously in quality and funding over the US. This is partly because different states want to do it differently (good), and partly because education is funded almost entirely out of local property taxes, so rich areas have super-schools, and poor areas have lousy schools (bad). The cycle reinforces as rich people move to areas with good schools.
I always read this, but at least here in NE Ohio it isn't true. Many of the better suburban schools spend significantly LESS per pupil than the Cleveland city schools do. Throwing money at schools won't fix the fact that the students don't want to learn and the parents don't have the will or ability to force them to. Sure there are a few utlra-wealthy suburbs that spend more than the big urban districts per pupil, but they don't really get better results than the ones like mine that just have a majority interested pupils who are provided the resources they need to learn.
Real example of competition in the wireless industry: my dad's in sales, he spend thousands of minutes on the phone every month. Back during the AMPs days he'd routinely have bills in the $300-400 range and he didn't use the phone nearly as much because of the expense. Today he has a family share plan with 5 lines and spends less than $200, which is even cheaper compared to the old rates due to inflation. I'd say competition is working quite well. The problem is dropping the cost of a single line because the telco has real costs related to billing that can't be reduced beyond a certain point.
Yeah my datacenter is 23-24C and the hottest disk bays in my SAN average about 37C. I don't care because my SAN is designed to lose an entire bay without losing data and the manufacturer is responsible for warranty replacement parts. So far in 22 months of operation it's lost three drive out of 160 and two of those were basically DOA with the other dying at about two months.
It's worse than your post implies because the manufacturers actually specify that drives be replaced every so often to get the MTBF rating. Basically the only thing an MTBF rating is good for is figuring out statistically what the chances are of a given RAID configuration losing data before a rebuild can be completed.
I would tend to agree with that. I run a datacenter that's cooled to 74 degrees and has good clean power from the online UPS's and I've had 6 drive failures out of about 500 drives over the last 22 months. Three were from older servers that weren't always properly cooled (the company had a crappy AC unit in their old data closet.) The other three all died in their first month or two after installation. So properly treated server class drives are dying at a rate of about.5% per year for me, I'd say that jives with manufacturer MTBF.
Bah, that's slow high latency memory. DDR2-667 CAS 2.5 is significantly more. For a workstation it might not matter, but for an 8 core DB server it does =)
Ok, it's come down a bit, but the kit for my HP boxes is still $775 per piece. I'd rather not fight with HP over warrant issues so I just use their branded memory.
4GB DIMM's are still extremely expensive, I know because we buy them for some servers. Generally it's significantly cheaper to buy two servers full of 2GB DIMM's then it is to fill one server with 4GB DIMM's ($800+ per 4GB stick vs $200 for 2GB). Of course you can buy workstations with 8 slots so that still gets you to 16GB.
Making a double or a triple album doesn't incur any kind of extra production cost.
Yeah, because studio time and studio engineers are free.... Sure your local garage band can put out a triple album of unlistenable crap, but even an unplugged album from your favorite artist is going to take some engineering time (unless the musician happens to also be a producer with a home recording studio, but that's pretty damn rare).
I use my QAM tuner. I get something like 50% of the HD channels my provider offers (all the locals) in HD without having to pay for a box or the HD pack. This saves me like $25/month. I don't know what the cable companies will are planning to do once they switch over to all digital, I would hope that they would support the basic tier over clear QAM so that you can continue to use a "cable ready" tv without having to pay to rent a box every month. If they don't I'll probably go to using OTA and online video rental.
$7,000 is the total purchase price per server. BBW (Battery Backed Write cache) is an option on almost all RAID cards, the base models come with RAID but not with BBW. 6Hr CTR, or 6 hour call to repair, it's the best service ever. It means that HP guarantees that they will have you back up and running in 6 hours, not parts/tech dispatched, but your problem actually resolved. When I call in and mention 6HR CTR all of the bullshit troubleshooting stops and they get to actually listening to my problem description and dispatching a technician. It costs about 8% of the purchase price for a 3 year contract, SO worth the money.
No, I included cooling. Even in the crappiest run of datacenters I can't imagine half of the electric load coming from UPS's and AC, but I included that in my calculations just to show how off theirs were.
You'd have to have a very cheap and very power inefficient server to come even remotely close to their claims of half of the cost of the server on power. An elbaso HP Dl360G5 costs $1600. It will use about 300W at typical load, but lets call it 250W to make the numbers easier. Double this for inefficient cooling and power conversion in the UPS (this is overly costly but makes up for underestimating power usage) so 500W. There are 8,760 hours in a year so 4,380 KwH, you'd have to pay $.20 per KwH to reach their figure, this is over twice the US national average. Prices where you'd want to put a datacenter are closer to $.06-$.08 per KwH. My average server cost closer to $7,000 with battery backed RAID card, dual fast drives, dual CPU's, 4GB memory, 3 year 6 hour repair contract, etc. Even powering that kind of servers off diesel generators fulltime it would have to draw ridiculous amounts of power to cost half it's purchase price in electricity every year.
In about 99% of cases they could migrate to OpenVMS/Itanium on modern hardware without a problem. Oh and saying scalability isn't an option about a VAX cluster is the silliest thing I think I've ever read on slashdot! VAX clusters make even an IBM sysplex look limiting!
Meh, the big UNIX boxes had plenty of I/O processors and bandwidth. The great reason to keep the mainframe around is JCL, because of JCL you can be assured that the job will complete in a given amount of time. Banks don't really care how fast a transaction completes, just that it will post by their deadline. It's best case vs average case vs worst case, UNIX and PC based servers can excel at the first two but absolutely suck at the last one, and that's why the mainframe is still around.
Slashdot Mirror
Yeah but those poles can generally hold the weight of another cable no problem and if you offer the right incentives to a municipality they will generally agree to enter into a franchise agreement with you. I know I personally use an overlay provider and love not having to deal with a monopoly provider (they actually refused to hook up my house because I had just moved in and didn't know where the grounding rod was for my electrical system!).
Yes but IAPV is also very common in non-collapsed hives so it is believed that it has to be a combination of factors with IAPV being but one (possibly major) factor.
You can definitly get the SAS drives in 15K, I use them for my Citrix blade servers (HP BL460C). During heavy login periods the 15K is a must, especially since the blade only allows two spindles.
They have 20% spare capacity which IS a lot more than you would put in a user servicable enclosure. The great thing is you get the speed advantage of the spares because while they limit the amount of data use to 80% they are normally using all spindles in the pack.
Can your homebrew solution scale from 1TB to 1.something Exabytes with 64GB of cache all without any downtime? Oh and does it come with a 5 year warranty? Also what do you do when your box is at a remote datacenter on another continent and you can't get ahold of an operator?
The broader issue is that primary and secondary education vary enormously in quality and funding over the US. This is partly because different states want to do it differently (good), and partly because education is funded almost entirely out of local property taxes, so rich areas have super-schools, and poor areas have lousy schools (bad). The cycle reinforces as rich people move to areas with good schools.
I always read this, but at least here in NE Ohio it isn't true. Many of the better suburban schools spend significantly LESS per pupil than the Cleveland city schools do. Throwing money at schools won't fix the fact that the students don't want to learn and the parents don't have the will or ability to force them to. Sure there are a few utlra-wealthy suburbs that spend more than the big urban districts per pupil, but they don't really get better results than the ones like mine that just have a majority interested pupils who are provided the resources they need to learn.
Real example of competition in the wireless industry: my dad's in sales, he spend thousands of minutes on the phone every month. Back during the AMPs days he'd routinely have bills in the $300-400 range and he didn't use the phone nearly as much because of the expense. Today he has a family share plan with 5 lines and spends less than $200, which is even cheaper compared to the old rates due to inflation. I'd say competition is working quite well. The problem is dropping the cost of a single line because the telco has real costs related to billing that can't be reduced beyond a certain point.
Yeah my datacenter is 23-24C and the hottest disk bays in my SAN average about 37C. I don't care because my SAN is designed to lose an entire bay without losing data and the manufacturer is responsible for warranty replacement parts. So far in 22 months of operation it's lost three drive out of 160 and two of those were basically DOA with the other dying at about two months.
It's worse than your post implies because the manufacturers actually specify that drives be replaced every so often to get the MTBF rating. Basically the only thing an MTBF rating is good for is figuring out statistically what the chances are of a given RAID configuration losing data before a rebuild can be completed.
We already fill em with 4GB DIMM's, it's just kind of expensive =)
I would tend to agree with that. I run a datacenter that's cooled to 74 degrees and has good clean power from the online UPS's and I've had 6 drive failures out of about 500 drives over the last 22 months. Three were from older servers that weren't always properly cooled (the company had a crappy AC unit in their old data closet.) The other three all died in their first month or two after installation. So properly treated server class drives are dying at a rate of about .5% per year for me, I'd say that jives with manufacturer MTBF.
Bah, that's slow high latency memory. DDR2-667 CAS 2.5 is significantly more. For a workstation it might not matter, but for an 8 core DB server it does =)
Ok, it's come down a bit, but the kit for my HP boxes is still $775 per piece. I'd rather not fight with HP over warrant issues so I just use their branded memory.
I don't think the GP was suggesting that the extra registers were AMD exclusive, more giving credit to AMD for the x64 design.
4GB DIMM's are still extremely expensive, I know because we buy them for some servers. Generally it's significantly cheaper to buy two servers full of 2GB DIMM's then it is to fill one server with 4GB DIMM's ($800+ per 4GB stick vs $200 for 2GB). Of course you can buy workstations with 8 slots so that still gets you to 16GB.
It's worse than that on Windows, unless you use the /3GB option in boot.ini 2GB is reserved for the kernel.
Doesn't even have to be modified, Getright has that feature built in, has since the 90's.
Making a double or a triple album doesn't incur any kind of extra production cost.
Yeah, because studio time and studio engineers are free.... Sure your local garage band can put out a triple album of unlistenable crap, but even an unplugged album from your favorite artist is going to take some engineering time (unless the musician happens to also be a producer with a home recording studio, but that's pretty damn rare).
I use my QAM tuner. I get something like 50% of the HD channels my provider offers (all the locals) in HD without having to pay for a box or the HD pack. This saves me like $25/month. I don't know what the cable companies will are planning to do once they switch over to all digital, I would hope that they would support the basic tier over clear QAM so that you can continue to use a "cable ready" tv without having to pay to rent a box every month. If they don't I'll probably go to using OTA and online video rental.
$7,000 is the total purchase price per server. BBW (Battery Backed Write cache) is an option on almost all RAID cards, the base models come with RAID but not with BBW. 6Hr CTR, or 6 hour call to repair, it's the best service ever. It means that HP guarantees that they will have you back up and running in 6 hours, not parts/tech dispatched, but your problem actually resolved. When I call in and mention 6HR CTR all of the bullshit troubleshooting stops and they get to actually listening to my problem description and dispatching a technician. It costs about 8% of the purchase price for a 3 year contract, SO worth the money.
No, I included cooling. Even in the crappiest run of datacenters I can't imagine half of the electric load coming from UPS's and AC, but I included that in my calculations just to show how off theirs were.
You'd have to have a very cheap and very power inefficient server to come even remotely close to their claims of half of the cost of the server on power. An elbaso HP Dl360G5 costs $1600. It will use about 300W at typical load, but lets call it 250W to make the numbers easier. Double this for inefficient cooling and power conversion in the UPS (this is overly costly but makes up for underestimating power usage) so 500W. There are 8,760 hours in a year so 4,380 KwH, you'd have to pay $.20 per KwH to reach their figure, this is over twice the US national average. Prices where you'd want to put a datacenter are closer to $.06-$.08 per KwH. My average server cost closer to $7,000 with battery backed RAID card, dual fast drives, dual CPU's, 4GB memory, 3 year 6 hour repair contract, etc. Even powering that kind of servers off diesel generators fulltime it would have to draw ridiculous amounts of power to cost half it's purchase price in electricity every year.
In about 99% of cases they could migrate to OpenVMS/Itanium on modern hardware without a problem. Oh and saying scalability isn't an option about a VAX cluster is the silliest thing I think I've ever read on slashdot! VAX clusters make even an IBM sysplex look limiting!
Actually they run Hyperion against Oracle on *gasp* a windows box.
Meh, the big UNIX boxes had plenty of I/O processors and bandwidth. The great reason to keep the mainframe around is JCL, because of JCL you can be assured that the job will complete in a given amount of time. Banks don't really care how fast a transaction completes, just that it will post by their deadline. It's best case vs average case vs worst case, UNIX and PC based servers can excel at the first two but absolutely suck at the last one, and that's why the mainframe is still around.