If I'm popping open 30-60 tabs in a single window, then I pretty much don't give a darn about the window titles at that point. Either I will utilize the drop-down listing of the tab titles (right end of the tab list in FF v3), or I'll simply [Ctrl-PgUp] and [Ctrl-PgDn] to work my way through them.
An example would be researching a product, browsing a web forum (opening up interesting topics in tabs), or other work where all of the tabs are rather similar in purpose.
If I truly want a separate group of tabs, then I'll open up a fresh browser window and use that window for the next collection of tabs.
If you plan on getting more than one, try to get all the same model. It is really convenient to not have to hunt for a specific power cord (or even a specific USB cord) if they all use the same cables.
Or go for the units that have built in PSUs and take a standard PC power cord.
It will definitely help with responsiveness. That is currently my biggest gripe with Firefox, where if I leave other windows open in the background, the content on those pages can steal CPU cycles. (Even with Flashblock and NoScript.)
The "abouttime" tag is very accurate. The question is whether it's a day late and a dollar short at this point. I've given serious thought to switching to Chrome.
- Laptops ordered in 2010/2011 will probably end up with Win7 by default.
- New systems ordered in 2010 will be purchased with the Win7 version that allows a downgrade to XP. We'll probably try out Win7 in that manner first.
- No old systems will be upgraded from XP to Win7.
We just spent 3 years upgrading everyone from Win98 to XP. Those boxes aren't slated to be retire for another 3-4 years at a minimum. Most are dual-core 2GB units, which should do fine until 2012 or later.
We skipped Vista and Vista SP1 entirely. But the crystal ball seems to indicate that Win7 won't be the complete dog that Vista was. And by the time we start upgrading again, the first or second service pack should be available. So we'll probably start the next upgrade cycle in 2012.
Or else Open Office will finally get a clue about usability and we'll switch to that (along with switching to Linux+WINE).
It's still worth doing, it cuts the rate of attacks by at least one order of magnitude (if not two) and your log files will be a lot cleaner. Yes, the worms are getting more sophisticated, but they also have to leave additional footprints in the IDS logs (because of having to port scan).
(We run public key only authentication, with SSH on a nonstandard port. The former is a bit of a PITA, but the latter is a simple change.)
When I moved up from PA to NY, I switched from a business class DSL line (1.5Mbps down, 384Mbps up) to Optimum Online Business. Back in PA, my DSL line used to run maxed out for weeks at a time (300 GB/month), up here I generally use anything between 100GB/mo up to 300-400GB/mo.
So far, in 2 years of being with Optimum Online Business, they have never hassled me about usage. Verizon/GTE never harassed me back in PA either about usage (in about 7 years of service).
The moral of the story is that if you want good service with no caps, pay for the business class service. That's usually in the $100-$150 range depending on the vendor and speeds. (I pay a bit less up here in NY then I did back in PA and get a lot more bandwidth.)
My experience with memtest is you can trust the results if it says the memory is bad, however if the memory passed it could still be bad. Troubleshooting your scenario should involve replacing the DIMM's in questions with known good modules while running Windows.
It is correct that MemTest86 (and MemTest86+) can only find truly broken DIMMs.
However, the better way to find timing issues or other problems that only occur when the system is under a load is to use burn-in tools. One favorite of the DIY crowd for over a decade is called Prime95 (from Mersenne.org). Because of the way that mersenne prime candidates are searched for, it places a heavy load on both the CPU and RAM. Which is ideal for uncovering issues with DIMMs that are not as fast as they claim to be.
In fact, because it was used so often for that purpose, the author(s) added a self-test / burn-in feature to the program.
For the most part, if you can run Prime95 for 48-72 hours straight with zero errors, then you're RAM/CPU are up-to-snuff and you're unlikely (barring a cosmic ray event) to encounter issues. You should also find a program to churn the hard drives during the test, in order to stress the system even more.
(On a multi-core system, you'll have to run multiple copies of Prime95. As of a few years ago, it was not yet multithreaded.)
The big advantage of hard drives for offsite storage is that they are a standard, self-contained unit, that can be attached to nearly any computer on the planet. They're also inexpensive, easy to grow (you don't have to replace a tape drive when you need the next size up), and very appealing to small businesses.
The downside is transportation costs along with bulk and no robotic mass handling of the units.
On the upside, if a solitary hard drive fails, you're only out the data on that drive. If a tape reader fails, it could take multiple tapes with it. Or, if you don't have a spare tape drive, you could be basically be stuck with oodles of tape that can't be read.
Tape makes sense once you reach the point where you have enough tapes to need automation. Where you're shipping data offsite every night in an express envelope or courier. Where you're locking away tapes every week/month for years at a time and the data isn't kept "live".
But for smaller shops with under 200-400GB worth of data, that is only growing a few dozen GB per year. Tape is simply overkill if not required by regulations. They're not interested in spending thousands on tape drives and thousands more on tape software and media. They'll be happy with half a dozen portable drives that they take offsite in a weekly rotation. Toss in a bit of nightly/weekly rsync (or rdiff-backup) to an offsite location and you're pretty much covered for a minimum of fuss.
Gradual is indeed the best way forward. Make sure that all newly purchased systems are completely legitimate and that all software going forward is also legitimate.
Eventually (a year or two), you'll get close to 100% legit, without breaking the bank.
At a minimum, you'll be making progress towards going legit.
(We also switched, over the course of a few years, as much software to OSS solutions as possible. With more switches planned.)
Check out eve online thats been doing the one universe thing for years. The universe isnt seamless since its split into a number of seperate systems but the entire population exists and plays within the same universe.
That only works for universes where one part of the game world is identical to nearly every other location in the game. Which, except for Jita CNAP 4-4, pretty much covers much of EVE's universe. It also helps that star systems in EVE Online are vast, so you could conceivably split a single star system across multiple hardware on the server side. The downside is that it makes the EVE universe feel bland and generic.
In fantasy MMOs, where locations are unique and different, you're going to end up with huge population concentrations. Lag is already bad enough when there are 50+ people sitting around the Ironforge Auction House or Naxx entry portal in WoW. Now, imagine that it's Saturday evening, and there are 500 or 5000 people in those locations. Your poor video card would pretty much melt into a pile of slag.
The 512 mb sd cards, OTOH, pitch 'em. I can't believe I'm saying this, but half a gig just isn't enough space to do anything with...movies are 700 Mb, as are most distros. (I use unetbootin to get away from burning CD's when testing out new distros these days.)
Useful for e-books (the Sony PRS-505 reader has an SD card slot). Or for use as floppy drives. Or for backups of things like encryption key rings and encrypted authentication credentials (basically a block of PGP/GPG encrypted text inside of text files, one for each website/company).
The big advantage is that they're super tiny in size, so you can toss a few in an envelope and store them in a safe, or safe deposit box.
There are lots of bit of information that can be usefully stored on 512MB sized cards (even down to 128MB), where it makes sense to keep the information spread across multiple pieces of physical media. (Such as rotational backups.)
Personally, I'm using 1TB drives from diff manufs (Samsungs seem to be a bit flaky) and waiting for the 2TB drives to be available from multiple manufs.
If you said 1GB, you'd be closer. One you add in anti-virus, a corporate instant message program, leaving e-mail up in the background all the time, plus a web browser and a few spreadsheet/documents, 512MB doesn't cut it.
But the better bet is still to spend a few extra dollars on the machine and get dual-core (more responsive machine) along with 2GB RAM (future-proofing).
Unless you want to be replacing machines still on a 3-year plan. (That extra 1GB of RAM and an extra core can easily make the machine into a 5-6 year box.)
1) Requires new hardware. VT is only available on newer Intel processors. So if you have an older server, and many people do, it isn't suitable for that purpose. That will become a non-issue eventually but at this time there are still lots of servers that aren't.
VT or HVM has been available for quite a while. All AM2 (and later) Athlon64 CPUs support hardware level virtualization, all Socket F Opterons support hardware level virtualization. (Both have been available for about 2-3 years now?)
The Intel side is murky for me, but I'd say that Intel has been in the HV game since mid-late 2007 as well. It's just harder to figure out which chips have it, and which companies haven't disabled it.
I'm still happy to see less expensive virtualization solutions. VMWare may "just work", but it's too damned expensive for small/mid sized shops with only a handful of physical servers.
BlueRay only requires 40 megabits/sec if you're using the old, outdated MPEG2 codec.
More modern codecs can give you that level of quality in anywhere from 1.5-3.0 megabits/sec. DVD quality can probably fit into a 768 kbit/sec stream, maybe as low as 600 kbit/sec.
MLC drives (with only a few tens of thousands of rewrite cycles) are cheap. Most thumbdrives are MLC, and all of the inexpensive SSDs are MLC.
SLC drives (with hundreds of thousands or millions of rewrite cycles) are expensive. These are basically the server-quality hardware (like IDE vs SCSI used to be).
That's the big difference in pricing. The other difference is the quality of the internal drive logic.
The major problem with this is that, if you guess wrong about how big a partition should be, it's a pain to resize things. So my usual thing is just to put/tmp on its own partition, and have a separate partition for / and for/home.
That's only a pain if you don't use LVM. Most of the Linux boxes that I setup have basically (4) partitions:
/boot - 256MB is overly generous. Useful to have on its own partition so you can keep it mounted as read-only.
/ - Anything from 8GB up to 32GB, and I can always shove any large trees elsewhere into LVM areas.
swap - Anything from 2GB up to 16GB. I should probably move this to LVM.
LVM area - this partition is where all of the other file systems live.
With LVM, it's about a 5 minute process to take a file system (at least for ext3) offline, enlarge the logical volume (LV), do the extend and fsck, and then bring the file system back online. LVM takes all of the guesswork out of initial sizing guesses. In fact, you're best to create numerous small file systems, then grow them as needed later.
Oh, I dunno. Get old enough and you start to want things to "just work". Which is something that Ubuntu gets mostly right for the desktop. Still some rough edges, but 8.04 was pretty nice.
Which is one of the things I ultimately didn't like about Gentoo. When you had a problem, first you had to figure out if it was due to bad USE flags, or whether it was a problem with the upstream package. Which made for a lot of fun fiddling, but it got tedious after a while.
Slashdot Mirror
That is incredibly dangerous to be outside for that length of time in such primitive suits.
I would not call today's suits primitive. Especially compared to the suits used during the Mercury/Apollo era.
Is there still room for improvement? Always. Massive improvements? Maybe not. Barring improvements in fabrics / construction technology.
If I'm popping open 30-60 tabs in a single window, then I pretty much don't give a darn about the window titles at that point. Either I will utilize the drop-down listing of the tab titles (right end of the tab list in FF v3), or I'll simply [Ctrl-PgUp] and [Ctrl-PgDn] to work my way through them.
An example would be researching a product, browsing a web forum (opening up interesting topics in tabs), or other work where all of the tabs are rather similar in purpose.
If I truly want a separate group of tabs, then I'll open up a fresh browser window and use that window for the next collection of tabs.
If you plan on getting more than one, try to get all the same model. It is really convenient to not have to hunt for a specific power cord (or even a specific USB cord) if they all use the same cables.
Or go for the units that have built in PSUs and take a standard PC power cord.
Rosewill RX82-U (JBOD)
BYTECC ME-835SU-SL
Lots of us *try* to do that with Firefox.
At least, until Firefox decides to stop responding as it spins its internal wheels.
Which pretty much makes "load in background" useless for its intended purpose.
It will definitely help with responsiveness. That is currently my biggest gripe with Firefox, where if I leave other windows open in the background, the content on those pages can steal CPU cycles. (Even with Flashblock and NoScript.) The "abouttime" tag is very accurate. The question is whether it's a day late and a dollar short at this point. I've given serious thought to switching to Chrome.
Our plan is basically:
- Laptops ordered in 2010/2011 will probably end up with Win7 by default.
- New systems ordered in 2010 will be purchased with the Win7 version that allows a downgrade to XP. We'll probably try out Win7 in that manner first.
- No old systems will be upgraded from XP to Win7.
We just spent 3 years upgrading everyone from Win98 to XP. Those boxes aren't slated to be retire for another 3-4 years at a minimum. Most are dual-core 2GB units, which should do fine until 2012 or later.
We skipped Vista and Vista SP1 entirely. But the crystal ball seems to indicate that Win7 won't be the complete dog that Vista was. And by the time we start upgrading again, the first or second service pack should be available. So we'll probably start the next upgrade cycle in 2012.
Or else Open Office will finally get a clue about usability and we'll switch to that (along with switching to Linux+WINE).
It's still worth doing, it cuts the rate of attacks by at least one order of magnitude (if not two) and your log files will be a lot cleaner. Yes, the worms are getting more sophisticated, but they also have to leave additional footprints in the IDS logs (because of having to port scan).
(We run public key only authentication, with SSH on a nonstandard port. The former is a bit of a PITA, but the latter is a simple change.)
When I moved up from PA to NY, I switched from a business class DSL line (1.5Mbps down, 384Mbps up) to Optimum Online Business. Back in PA, my DSL line used to run maxed out for weeks at a time (300 GB/month), up here I generally use anything between 100GB/mo up to 300-400GB/mo.
So far, in 2 years of being with Optimum Online Business, they have never hassled me about usage. Verizon/GTE never harassed me back in PA either about usage (in about 7 years of service).
The moral of the story is that if you want good service with no caps, pay for the business class service. That's usually in the $100-$150 range depending on the vendor and speeds. (I pay a bit less up here in NY then I did back in PA and get a lot more bandwidth.)
Or you could have told PostgreSQL to do a case-insensitive search on that SQL statement.
See [NOVICE] Case Insensitive Searching?.
My experience with memtest is you can trust the results if it says the memory is bad, however if the memory passed it could still be bad. Troubleshooting your scenario should involve replacing the DIMM's in questions with known good modules while running Windows.
It is correct that MemTest86 (and MemTest86+) can only find truly broken DIMMs.
However, the better way to find timing issues or other problems that only occur when the system is under a load is to use burn-in tools. One favorite of the DIY crowd for over a decade is called Prime95 (from Mersenne.org). Because of the way that mersenne prime candidates are searched for, it places a heavy load on both the CPU and RAM. Which is ideal for uncovering issues with DIMMs that are not as fast as they claim to be.
In fact, because it was used so often for that purpose, the author(s) added a self-test / burn-in feature to the program.
For the most part, if you can run Prime95 for 48-72 hours straight with zero errors, then you're RAM/CPU are up-to-snuff and you're unlikely (barring a cosmic ray event) to encounter issues. You should also find a program to churn the hard drives during the test, in order to stress the system even more.
(On a multi-core system, you'll have to run multiple copies of Prime95. As of a few years ago, it was not yet multithreaded.)
The big advantage of hard drives for offsite storage is that they are a standard, self-contained unit, that can be attached to nearly any computer on the planet. They're also inexpensive, easy to grow (you don't have to replace a tape drive when you need the next size up), and very appealing to small businesses.
The downside is transportation costs along with bulk and no robotic mass handling of the units.
On the upside, if a solitary hard drive fails, you're only out the data on that drive. If a tape reader fails, it could take multiple tapes with it. Or, if you don't have a spare tape drive, you could be basically be stuck with oodles of tape that can't be read.
Tape makes sense once you reach the point where you have enough tapes to need automation. Where you're shipping data offsite every night in an express envelope or courier. Where you're locking away tapes every week/month for years at a time and the data isn't kept "live".
But for smaller shops with under 200-400GB worth of data, that is only growing a few dozen GB per year. Tape is simply overkill if not required by regulations. They're not interested in spending thousands on tape drives and thousands more on tape software and media. They'll be happy with half a dozen portable drives that they take offsite in a weekly rotation. Toss in a bit of nightly/weekly rsync (or rdiff-backup) to an offsite location and you're pretty much covered for a minimum of fuss.
Gradual is indeed the best way forward. Make sure that all newly purchased systems are completely legitimate and that all software going forward is also legitimate.
Eventually (a year or two), you'll get close to 100% legit, without breaking the bank.
At a minimum, you'll be making progress towards going legit.
(We also switched, over the course of a few years, as much software to OSS solutions as possible. With more switches planned.)
Start with Linux on an unimportant portion of your systems, but one that you'll be using regularly.
For us, it was an SVN server. And at home, a Samba file server.
Once we mastered that, we got into Xen and virtualization.
Then we replaced our firewall with a Linux box, running Squid.
Then we moved all of our mail server duties over to a Linux box. And setup PostgreSQL on Linux for our database needs.
Our next goal is to remove the Windows file servers. And to switch over to Apache using either PHP or Java instead of IIS+VBScript.
Check out eve online thats been doing the one universe thing for years. The universe isnt seamless since its split into a number of seperate systems but the entire population exists and plays within the same universe.
That only works for universes where one part of the game world is identical to nearly every other location in the game. Which, except for Jita CNAP 4-4, pretty much covers much of EVE's universe. It also helps that star systems in EVE Online are vast, so you could conceivably split a single star system across multiple hardware on the server side. The downside is that it makes the EVE universe feel bland and generic.
In fantasy MMOs, where locations are unique and different, you're going to end up with huge population concentrations. Lag is already bad enough when there are 50+ people sitting around the Ironforge Auction House or Naxx entry portal in WoW. Now, imagine that it's Saturday evening, and there are 500 or 5000 people in those locations. Your poor video card would pretty much melt into a pile of slag.
If you run udev, you can often get an unique identifier by looking at:
/dev/disk/by-id/ ../../sdp
# ls -l
usb-ST950032_5AS_ST9500325A_5QZ07NVA ->
This allows you to uniquely identify USB drives, and you can even setup udev to do auto-mounting in order to run backup scripts.
Just another trick when using USB drives.
The 512 mb sd cards, OTOH, pitch 'em. I can't believe I'm saying this, but half a gig just isn't enough space to do anything with...movies are 700 Mb, as are most distros. (I use unetbootin to get away from burning CD's when testing out new distros these days.)
Useful for e-books (the Sony PRS-505 reader has an SD card slot). Or for use as floppy drives. Or for backups of things like encryption key rings and encrypted authentication credentials (basically a block of PGP/GPG encrypted text inside of text files, one for each website/company).
The big advantage is that they're super tiny in size, so you can toss a few in an envelope and store them in a safe, or safe deposit box.
There are lots of bit of information that can be usefully stored on 512MB sized cards (even down to 128MB), where it makes sense to keep the information spread across multiple pieces of physical media. (Such as rotational backups.)
Prices on 1.5TB have dropped like a rock. They are, in units/dollar, cheaper than .5TB, slightly cheaper than 1TB, and roughly equal to .75TB.
You mean the 1.5TB Seagates which are now notorious for being flaky drives?
Personally, I'm using 1TB drives from diff manufs (Samsungs seem to be a bit flaky) and waiting for the 2TB drives to be available from multiple manufs.
If you said 1GB, you'd be closer. One you add in anti-virus, a corporate instant message program, leaving e-mail up in the background all the time, plus a web browser and a few spreadsheet/documents, 512MB doesn't cut it.
But the better bet is still to spend a few extra dollars on the machine and get dual-core (more responsive machine) along with 2GB RAM (future-proofing).
Unless you want to be replacing machines still on a 3-year plan. (That extra 1GB of RAM and an extra core can easily make the machine into a 5-6 year box.)
1) Requires new hardware. VT is only available on newer Intel processors. So if you have an older server, and many people do, it isn't suitable for that purpose. That will become a non-issue eventually but at this time there are still lots of servers that aren't.
VT or HVM has been available for quite a while. All AM2 (and later) Athlon64 CPUs support hardware level virtualization, all Socket F Opterons support hardware level virtualization. (Both have been available for about 2-3 years now?)
The Intel side is murky for me, but I'd say that Intel has been in the HV game since mid-late 2007 as well. It's just harder to figure out which chips have it, and which companies haven't disabled it.
I'm still happy to see less expensive virtualization solutions. VMWare may "just work", but it's too damned expensive for small/mid sized shops with only a handful of physical servers.
BlueRay only requires 40 megabits/sec if you're using the old, outdated MPEG2 codec.
More modern codecs can give you that level of quality in anywhere from 1.5-3.0 megabits/sec. DVD quality can probably fit into a 768 kbit/sec stream, maybe as low as 600 kbit/sec.
A bigger problem is our reluctance to move off 512-byte sectors. Who needs that fine granularity of LBA?
/. front page in '07 or '08.
There are already plans afoot to move to a 4096 byte sector size. They started a few years ago, and might have even made
MLC drives (with only a few tens of thousands of rewrite cycles) are cheap. Most thumbdrives are MLC, and all of the inexpensive SSDs are MLC.
SLC drives (with hundreds of thousands or millions of rewrite cycles) are expensive. These are basically the server-quality hardware (like IDE vs SCSI used to be).
That's the big difference in pricing. The other difference is the quality of the internal drive logic.
The major problem with this is that, if you guess wrong about how big a partition should be, it's a pain to resize things. So my usual thing is just to put /tmp on its own partition, and have a separate partition for / and for /home.
/boot - 256MB is overly generous. Useful to have on its own partition so you can keep it mounted as read-only.
That's only a pain if you don't use LVM. Most of the Linux boxes that I setup have basically (4) partitions:
/ - Anything from 8GB up to 32GB, and I can always shove any large trees elsewhere into LVM areas.
swap - Anything from 2GB up to 16GB. I should probably move this to LVM.
LVM area - this partition is where all of the other file systems live.
With LVM, it's about a 5 minute process to take a file system (at least for ext3) offline, enlarge the logical volume (LV), do the extend and fsck, and then bring the file system back online. LVM takes all of the guesswork out of initial sizing guesses. In fact, you're best to create numerous small file systems, then grow them as needed later.
Oh, I dunno. Get old enough and you start to want things to "just work". Which is something that Ubuntu gets mostly right for the desktop. Still some rough edges, but 8.04 was pretty nice.
Which is one of the things I ultimately didn't like about Gentoo. When you had a problem, first you had to figure out if it was due to bad USE flags, or whether it was a problem with the upstream package. Which made for a lot of fun fiddling, but it got tedious after a while.
As far as I can tell, there is no more actual metamoderation.
There is, but it no longer shows up on the front page. Sometimes when I submit a comment, I then get offered to meta-mod.