Not to mention that everyone selling 4 way and larger x64 servers offers raided memory if you want it. My biggest gripe with x64 systems is the lack of sufficient I/O offloading. High workloads are fairly easily met by the CPU and memory subsystems but when it comes to moving big piles of data to and from the network and storage they kind of suck. We get fairly good performance by pinning our big database tables in memory and by using TOE cards (which are poorly supported) for networking. There is some hope on the horizon with many 10gig ethernet adapters being CNA's with a high degree of offloading, but it's one area where I think the x64 market needs to mature a bit more.
Hahaha, Nehalem is the cheapest virtualization platform by a LARGE margin. The ability to do greater than 2 DIMM's per core is huge. Unless you are talking about HPC in which case the balance of performance, price, memory and power usage also generally comes out in favor of Nehalem.
Probably too much money to buy insurance, if you're that large having multiple physically dispersed systems IS your insurance. For the customers the insurance is included in the cost of the service, Amazon adds a little bit to the bill and puts it in a fund in case they have to issue SLA related refunds (or more likely they just make less that month due to issuing credits).
No, MS will not enter into agreements with OEM's, that's what got them in trouble in the first place! They will have the kit available to OEM's but I really think they will stay far away from an official or unofficial pressure to use it. For consumers they will probably offer it as an optional component in Windows Update which hasn't been tied to IE since Vista launched.
They replaced them with mirrored SRAM CPU's. The real issue was one of poor design, they had ECC throughout the system including system busses but had only simple parity checks on the SRAM. By having essentially RAID1 on the SRAM they could recover from most flipped bit problems, the exception being writeback cache errors that had already pushed bad data to main ram (silent data corruption). Those were much less common than data cache problems that caused system halts and could have their impact mitigated by turning up the cache scrubbing interval.
Here's an article talking about the software "patch" that Sun came up with before finally admitting they had bad parts. The company I supported at the time ended up having thousands of the processors replaced with parts with redundant SRAM modules from whatever their best supplier was. They were literally getting multiple system halts a day across their enterprise caused by the bad cache SRAM. Sun tried blaming it on cosmic rays and some other junk before finally issuing a replacement order under their enterprise support agreement.
I'd suggest Livelink by OpenText. I know the Airforce uses it since our Livelink guy worked on their systems before coming to work for us, they obviously work with large volumes of aerospace related documents! =) That probably means OpenText can find consultants who have already designed and worked with an aerospace taxonomy.
HP has a SFF 15K drive with a sub 5ms full stroke access time (part 512547-B21). I think it's probably a 1.8 or 2" platter. Very good if you are worried about worst case performance for DB work. Oh and the RAID penalty you mention is of course why Oracle says SAME =)
They looked at the EV-1 as a solution to a legislative (not economic) problem. Once they got California to back down on the zero emission requirement and bought federal laws that said noone could be more restrictive than California they figured there was little need to keep the program around. Since 51+% of passenger vehicles sold were light trucks and SUV's I would say their reasoning was fairly sound.
If you're old enough I'm sure you did through the old AT&T. Bell labs might as well have been publicly owned since it was created by government mandate and supported through backdoor taxes of monopoly pricing.
Actually it doesn't mark it for deletion because the block may still be part of a snapshots filesystem bitmap. In this way it's quite similar to WAFL from Netapp. What the cleanup process does is checks the currently allocated block bitmap to all of other bitmaps and frees any allocated blocks that aren't in use by anything.
I want full detail at 1080P60 which is the native resolution of my LCD. Unfortunately I don't want to pay for either the components or the power to achieve that. I get closer to 1080P30 with an Athlon x2 4200+HE and 9600GSO, my whole system uses about a third the power of just the graphics cards in a top end system today.
Electric potential has almost zero chance of taking down an airliner because they are not grounded, low level turbulance and downdrafts on the other hand have a long history of taking out aircraft.
AT&T DOES allow tethering, just not with the iPhone plan. We added tethering to our corporate BB plans recently as we are upgrading to Bold's (3G) and it's cheaper than having a bunch of laptop only contracts.
Obviously, MOST people and corporations moving freight find that rail and truck are both more economical than air - witness the fact that millions of tons of freight roll down the tracks and the highways each and every night, whereas air freight is reserved for small, high priority shipments. (In fact, shipping by truck is often faster than shipping by air, but I won't go into that here)
You'd be surprised at how much heavy stuff goes air cargo. We routinely have servers shipped air freight for instance. Ford at one point in the 90's was selling cars so fast that they were building engines in Cleveland and flying them across Lake Erie to the assembly plant in Canada because it was cheaper to fly the engine than it was to lose the sales. That of course is taking JIT manufacturing to the extreme and a buffer of engines at the assembly plant might have been preferable to allow more efficient transportation of new engines, but there are definitely carrying costs to having that extra inventory and the storage space for it.
Slashdot Mirror
Not to mention that everyone selling 4 way and larger x64 servers offers raided memory if you want it. My biggest gripe with x64 systems is the lack of sufficient I/O offloading. High workloads are fairly easily met by the CPU and memory subsystems but when it comes to moving big piles of data to and from the network and storage they kind of suck. We get fairly good performance by pinning our big database tables in memory and by using TOE cards (which are poorly supported) for networking. There is some hope on the horizon with many 10gig ethernet adapters being CNA's with a high degree of offloading, but it's one area where I think the x64 market needs to mature a bit more.
Hahaha, Nehalem is the cheapest virtualization platform by a LARGE margin. The ability to do greater than 2 DIMM's per core is huge. Unless you are talking about HPC in which case the balance of performance, price, memory and power usage also generally comes out in favor of Nehalem.
Probably too much money to buy insurance, if you're that large having multiple physically dispersed systems IS your insurance. For the customers the insurance is included in the cost of the service, Amazon adds a little bit to the bill and puts it in a fund in case they have to issue SLA related refunds (or more likely they just make less that month due to issuing credits).
Uh, tell that to the people who lost loved ones in Bali, I'd say there are plenty of radical Muslim terrorists in SE Asia.
That's what HP does with iLo passwords, the unique default password is printed on the toe-tag of each server.
Actually HP's b and c class blade servers had/have a full enclosure for each blade.
No, MS will not enter into agreements with OEM's, that's what got them in trouble in the first place! They will have the kit available to OEM's but I really think they will stay far away from an official or unofficial pressure to use it. For consumers they will probably offer it as an optional component in Windows Update which hasn't been tied to IE since Vista launched.
They replaced them with mirrored SRAM CPU's. The real issue was one of poor design, they had ECC throughout the system including system busses but had only simple parity checks on the SRAM. By having essentially RAID1 on the SRAM they could recover from most flipped bit problems, the exception being writeback cache errors that had already pushed bad data to main ram (silent data corruption). Those were much less common than data cache problems that caused system halts and could have their impact mitigated by turning up the cache scrubbing interval.
Here's an article talking about the software "patch" that Sun came up with before finally admitting they had bad parts. The company I supported at the time ended up having thousands of the processors replaced with parts with redundant SRAM modules from whatever their best supplier was. They were literally getting multiple system halts a day across their enterprise caused by the bad cache SRAM. Sun tried blaming it on cosmic rays and some other junk before finally issuing a replacement order under their enterprise support agreement.
It's pretty damn expensive to pay a shill and incredibly hard to pass card information to the shill without detection.
Online you can't know that the house isn't running a bot allowed to peak at the deck or your hand.
Enterprise Content Management.
I'd suggest Livelink by OpenText. I know the Airforce uses it since our Livelink guy worked on their systems before coming to work for us, they obviously work with large volumes of aerospace related documents! =) That probably means OpenText can find consultants who have already designed and worked with an aerospace taxonomy.
What you described can be done with any decent raid controller.
HP has a SFF 15K drive with a sub 5ms full stroke access time (part 512547-B21). I think it's probably a 1.8 or 2" platter. Very good if you are worried about worst case performance for DB work. Oh and the RAID penalty you mention is of course why Oracle says SAME =)
UDF should also be worth a try. Vista and current versions of OSX support it read/write and it should support all the features you list no problem.
They looked at the EV-1 as a solution to a legislative (not economic) problem. Once they got California to back down on the zero emission requirement and bought federal laws that said noone could be more restrictive than California they figured there was little need to keep the program around. Since 51+% of passenger vehicles sold were light trucks and SUV's I would say their reasoning was fairly sound.
If you're old enough I'm sure you did through the old AT&T. Bell labs might as well have been publicly owned since it was created by government mandate and supported through backdoor taxes of monopoly pricing.
Oh god the Ultra II 450 cache bug, that one was annoying as hell.
Actually it doesn't mark it for deletion because the block may still be part of a snapshots filesystem bitmap. In this way it's quite similar to WAFL from Netapp. What the cleanup process does is checks the currently allocated block bitmap to all of other bitmaps and frees any allocated blocks that aren't in use by anything.
I want full detail at 1080P60 which is the native resolution of my LCD. Unfortunately I don't want to pay for either the components or the power to achieve that. I get closer to 1080P30 with an Athlon x2 4200+HE and 9600GSO, my whole system uses about a third the power of just the graphics cards in a top end system today.
Electric potential has almost zero chance of taking down an airliner because they are not grounded, low level turbulance and downdrafts on the other hand have a long history of taking out aircraft.
AT&T DOES allow tethering, just not with the iPhone plan. We added tethering to our corporate BB plans recently as we are upgrading to Bold's (3G) and it's cheaper than having a bunch of laptop only contracts.
MONEY, they get some (large) percentage of the monthly contract for exclusivity.
Obviously, MOST people and corporations moving freight find that rail and truck are both more economical than air - witness the fact that millions of tons of freight roll down the tracks and the highways each and every night, whereas air freight is reserved for small, high priority shipments. (In fact, shipping by truck is often faster than shipping by air, but I won't go into that here)
You'd be surprised at how much heavy stuff goes air cargo. We routinely have servers shipped air freight for instance. Ford at one point in the 90's was selling cars so fast that they were building engines in Cleveland and flying them across Lake Erie to the assembly plant in Canada because it was cheaper to fly the engine than it was to lose the sales. That of course is taking JIT manufacturing to the extreme and a buffer of engines at the assembly plant might have been preferable to allow more efficient transportation of new engines, but there are definitely carrying costs to having that extra inventory and the storage space for it.