The only problems I've had is how in relationship to certain HP printers using Postscript drivers, they'll occasionally run into problems with certain PDF's containing values that are valid in later PDF specs but not valid for PS. If we hadn't done a huge amount of form design around PS I'd switch em to PCL and be done with it.
That only works because someone else has already depreciated the asset, *someone* has to pay full price to keep the R&D and production facilities running. I'm not saying that essentially unlimited budgets like Avatar have to be the norm, but the real average per film is going to be well north of $7k.
Reed Timmers group (seen on Storm Chasers on Discovery) did exactly that, they had a large model plane built with an ejection system that drops parachute probes into the inflow current of the storm where they are then lifted and circulated near the core. They also use synthetic aperture radar on an armored vehicle to get it very close or into the storm, this has the potential to grab significantly more information than a swarm of DOW vehicles at large standoff distances being obstructed by the rain curtain that wraps most tornado's.
The best thing about Redbox if you have kids and do roadtrips is you can rent at one location and return at another. When we were driving back from Florida to Ohio and the kids had already watched all the DVD's we brought we were able to rent 10 new ones and return them when we got home. It was probably the best $10 spent in the history of humanity =)
Big movies also tend to have a TON of special effects and CG which burn computer time like it's going out of style and use armies of fairly well compensated graphics and computer professionals. Also the salaries of the stars and the advertising budget are both significant contributions. The actual cost for materials and such is probably only 4-5x that of an independent film.
Yeah, I'm thinking long term more useful data is going to come out of small teams like Reed Timmers using remote probes and synthetic aperture radar near to the storm in armored vehicles then having hundreds of guys collecting wind speed and velocity data around the storm at ground level and using DOW trucks from outside the storm.
Looks like material costs would be similar per area to collecting solar plants but with a MUCH lower efficiency and without the ability to store energy ala molten salt solar collectors. The only advantage they have is the greenhouse effect, but most places likely to house solar plants don't have enough water for that to be a major consideration.
Nope, that's 2TB/day across 20GB (I believe, logged off my corp systems a while ago but it's in the low 10's of GB regardless of the actual size). It's the redo log volumes for a fairly high transaction load OLTP Oracle server.
I was referring to larger in the sense of writes per day, they are only a few tens of GB each, they are the redo log volumes for our OLTP Oracle servers.
It totally depends on the use case. Some of my larger san volumes show 2TB/day of writes which means according to Intel's x-25e datasheet a 64GB drive would last ~1,000 days or under 3 years.
Nope, just use the more powerful servers and up the consolidation ratio at night and power off unneeded hosts. VMWare supports just this with vSphere, it's called DPM or dynamic power management.
Which would you rather have in 8kw, 2048 atom cores or 768 56xx cores? Because that's what HP can cram into 4xc7000's in the same power and floor space envelope (not sure about capital outlay though). Personally for my workloads I'd take the 56xx cores, but YMMV.
Think TAR backing up a directory while preserving ACL's and change/create/access dates as well as file attributes and putting all that information into a database, because that and optionally streaming the data over the network to a backup server is the basics of what Netbackup does. It can do raw disk backups with metadata on windows but I'm not sure I trust it to work correctly and not mess up the server plus it requires an additional license.
We've had two failures in 4200+ tapes in the last 4 years using mostly LTO3 and one of those was dropped so I don't count it. Were you really seeing significantly higher failure rates?!?
Try 70 drives in 5U (HP SW 500), but that doesn't negate the fact that HDD's sitting in an array are not a backup, just a copy. If it's not offline, offsite, and independently redundant it's not a backup =) Oh and on cost I can get twice as much storage as the $50k fully loaded SW 500 for under $20k with LTO4 (in fact I did last month).
Yes, I was assuming his experience with tape was QIC/DAT/Traven/etc based. I've been using real enterprise class tape for almost 15 years and I've had problems with the hardware a handful of times, my experience has been that the software is usually the major problem in the backup space. Heck in my "old" LTO3 library I've put over 5k tapes through the thing and lost one drive in 4 years and had two bad tapes, one of which was dropped onto the tile floor. On the software side, as much as I hate Symantec, Netbackup is still the best of the bad products I've used with Commvault a close second. I've never worked in an environment with TSM so I can't comment on its level of suck, but I do know that it's expensive enough to make Netbackup look cheap which is really saying something.
Yep, but that's what it takes to keep it running above minimum speed, writing a single large filesystem volume takes longer than writing 4 volumes almost as large because the drive shoeshines with a single job. Minimum rate for the drive is 40MB/s, the best I have done with tuning on a 72 drive vraid6 volume is 35MB/s and average is closer to 25MB/s sustained but 4 jobs from the same array on different volumes will give me 100-120MB/s. All volumes are spanned across all disks so it's not a matter of more spindles being available, it's the latency in all the metadata lookups.
Must be using cheap consumer based tape, DLT, LTO, and IBM's various tape formats are solid, reliable, and available. I can still buy new drives today that will read tapes written 20+ years ago and media 4 generations old is still available new.
Slashdot Mirror
The only problems I've had is how in relationship to certain HP printers using Postscript drivers, they'll occasionally run into problems with certain PDF's containing values that are valid in later PDF specs but not valid for PS. If we hadn't done a huge amount of form design around PS I'd switch em to PCL and be done with it.
Sorry, but I don't want a reset clearing any user data on persistent storage.
That only works because someone else has already depreciated the asset, *someone* has to pay full price to keep the R&D and production facilities running. I'm not saying that essentially unlimited budgets like Avatar have to be the norm, but the real average per film is going to be well north of $7k.
Reed Timmers group (seen on Storm Chasers on Discovery) did exactly that, they had a large model plane built with an ejection system that drops parachute probes into the inflow current of the storm where they are then lifted and circulated near the core. They also use synthetic aperture radar on an armored vehicle to get it very close or into the storm, this has the potential to grab significantly more information than a swarm of DOW vehicles at large standoff distances being obstructed by the rain curtain that wraps most tornado's.
The best thing about Redbox if you have kids and do roadtrips is you can rent at one location and return at another. When we were driving back from Florida to Ohio and the kids had already watched all the DVD's we brought we were able to rent 10 new ones and return them when we got home. It was probably the best $10 spent in the history of humanity =)
enough to put the bar for entry at around $1500, give or take.
I don't think you can buy the glass for a decent film camera for less than $1500 let alone the film or digital camera itself.
Big movies also tend to have a TON of special effects and CG which burn computer time like it's going out of style and use armies of fairly well compensated graphics and computer professionals. Also the salaries of the stars and the advertising budget are both significant contributions. The actual cost for materials and such is probably only 4-5x that of an independent film.
Yeah, I'm thinking long term more useful data is going to come out of small teams like Reed Timmers using remote probes and synthetic aperture radar near to the storm in armored vehicles then having hundreds of guys collecting wind speed and velocity data around the storm at ground level and using DOW trucks from outside the storm.
Looks like material costs would be similar per area to collecting solar plants but with a MUCH lower efficiency and without the ability to store energy ala molten salt solar collectors. The only advantage they have is the greenhouse effect, but most places likely to house solar plants don't have enough water for that to be a major consideration.
We lose about 1.5-2% of spindles per year, losing effectively 33+% per year of already expensive spindles doesn't really work out too well.
Nope, that's 2TB/day across 20GB (I believe, logged off my corp systems a while ago but it's in the low 10's of GB regardless of the actual size). It's the redo log volumes for a fairly high transaction load OLTP Oracle server.
I was referring to larger in the sense of writes per day, they are only a few tens of GB each, they are the redo log volumes for our OLTP Oracle servers.
It totally depends on the use case. Some of my larger san volumes show 2TB/day of writes which means according to Intel's x-25e datasheet a 64GB drive would last ~1,000 days or under 3 years.
Or have a good home theater that uses the sound to put you into the stadium =)
Latest generation Intel Xeon CPU's.
Nope, just use the more powerful servers and up the consolidation ratio at night and power off unneeded hosts. VMWare supports just this with vSphere, it's called DPM or dynamic power management.
Which would you rather have in 8kw, 2048 atom cores or 768 56xx cores? Because that's what HP can cram into 4xc7000's in the same power and floor space envelope (not sure about capital outlay though). Personally for my workloads I'd take the 56xx cores, but YMMV.
You're missing the point, you still have to pull the metadata if you want a consistent backup and to be able to do differentials.
Think TAR backing up a directory while preserving ACL's and change/create/access dates as well as file attributes and putting all that information into a database, because that and optionally streaming the data over the network to a backup server is the basics of what Netbackup does. It can do raw disk backups with metadata on windows but I'm not sure I trust it to work correctly and not mess up the server plus it requires an additional license.
We've had two failures in 4200+ tapes in the last 4 years using mostly LTO3 and one of those was dropped so I don't count it. Were you really seeing significantly higher failure rates?!?
Try 70 drives in 5U (HP SW 500), but that doesn't negate the fact that HDD's sitting in an array are not a backup, just a copy. If it's not offline, offsite, and independently redundant it's not a backup =) Oh and on cost I can get twice as much storage as the $50k fully loaded SW 500 for under $20k with LTO4 (in fact I did last month).
linear read != real file server backup.
Yes, I was assuming his experience with tape was QIC/DAT/Traven/etc based. I've been using real enterprise class tape for almost 15 years and I've had problems with the hardware a handful of times, my experience has been that the software is usually the major problem in the backup space. Heck in my "old" LTO3 library I've put over 5k tapes through the thing and lost one drive in 4 years and had two bad tapes, one of which was dropped onto the tile floor. On the software side, as much as I hate Symantec, Netbackup is still the best of the bad products I've used with Commvault a close second. I've never worked in an environment with TSM so I can't comment on its level of suck, but I do know that it's expensive enough to make Netbackup look cheap which is really saying something.
Yep, but that's what it takes to keep it running above minimum speed, writing a single large filesystem volume takes longer than writing 4 volumes almost as large because the drive shoeshines with a single job. Minimum rate for the drive is 40MB/s, the best I have done with tuning on a 72 drive vraid6 volume is 35MB/s and average is closer to 25MB/s sustained but 4 jobs from the same array on different volumes will give me 100-120MB/s. All volumes are spanned across all disks so it's not a matter of more spindles being available, it's the latency in all the metadata lookups.
Must be using cheap consumer based tape, DLT, LTO, and IBM's various tape formats are solid, reliable, and available. I can still buy new drives today that will read tapes written 20+ years ago and media 4 generations old is still available new.