If you are going to do virtualization, the only benefit comes when you invest in a cluster otherwise don't do it at all.
This is not true at all. Indeed, the benefits of virtualisation are such that even for a single service on a single server, it's generally better to make it a VM.
If you decide to go that route regardless, invest in at least 2 dual-port 4GB Fiber HBAs per host server. You'll be thankful in the long run.
Do you have any idea how much disk you need on the back end to exceed the throughput of even a single-port 4Gb HBA, for more than a second or two ?
There are a lot of places to spend money before worrying about how fat your server-side SAN connections are. A *LOT*. 99% of the time, a single 1Gb iSCSI port is more than enough from a performance perspective.
If your take-home pay is >$42,000 a month (very conservative estimate for 1M/year gross), do you really need a car loan?
A "six figure income" starts at $100,000, not $1,000,000.
And if you're in that situation, and you do need a loan, that doesn't speak well for your money management skills.
With low financing rates like 0% - 2% being quite common, I'd be mad *not* to take advantage of it and use the cash somewhere it would get a better return.
We don't use cheques, we use checks. I don't particularly care that you need to re-enforce the differences our languages have developed over the years, most of us are aware that we spell somethings different just so its clear we are NOT YOU. Get over yourself.
You're the only person here who feels the need to "re-enforce the differences". I write it that way because that's the way I've been writing it for thirty years.
I bought my first car with a low interest rate and the car cost more than half my yearly salary, it was the first thing I bought on credit. Not really sure what your problem is, but until about a year ago having no credit wasn't really an issue for any one unless you were old enough that no credit meant you had probably had bad credit and waited for the bad stuff to disappear off your record.
The problem is when you arrive in the USA, you have no credit. The ridiculous part is where this is inescapably equated with having bad credit.
I can't believe you people still haven't gotten over yourselves. Keep it up, at least the rest of the world can see where we got our arrogance from that way. By the way, hows' that world power thing going for you now days?
With a "6 figure" income you need to borrow money to buy basic transportation?
No, I just know that my money can be put to better use than buying a car with cash or sitting in deposit for a secured loan. Particularly in the very expensive first few months of an international move (and especially before I receive my first paycheque).
Can I have a $30,000 loan? I have a six-figure income, I can even have you meet my employer to prove it. What's that? You don't like giving loans to people you just meet, even if they make $120k a year?
Yet in many countries (Australia, Switzerland, England, to name the ones I am personally familiar with) you can do just that - simple proof of employment and income is all that's necessary (unless you're self-employed). I had a credit card with a $20k limit within a couple of months of arriving in Switzerland, and the only reason it took that long was the wait for my residency permit to be processed. All they wanted was evidence that I was a legal resident and proof of income. Further, stupidities like a "credit check" with too large an outstanding amount on a card resulting in a loss of "credit rating" were not even a passing concern.
Really, the biggest problem dealing with US banks as a new resident is that anything not American may as well not exist. Five year history with same company ? Tens of thousands of dollars in cash back home ? Existing mortgage with a decade of regular repayments and rental income ? Several existing credit cards ? If you're new to the country it all means diddly squat. You're in the same boat as some random kid just finishing high school who's never had a job - you have to spend 6 months (if you're lucky) to 12 months (more commonly) walking on eggshells. It's ludicrous.
And we didn't fall for that chip-and-PIN-and-you're-liable-if-your-card-is-lost scam.
What "scam" ? I'm not aware of any countries where the customer is liable for fraudulent transactions unless they've been grossly negligent (and being grossly negligent with your "checks" won't produce any different result).
Anyhow, taking a loan to buy a car is almost always a mistake. Save your money, pay cash, be free.
There are numerous situations where it makes more sense to get a loan. Everything from not having sufficient cash on hand, to being able to get a better return on the money elsewhere.
Citation please. I was able to qualify for a low (0%) loan on a new vehicle, with only your normal everyday 5 figure income. Dodge Diesel truck--not some econobox.
Because - as I have recently found after starting a job here - the American banking system is utterly insane.
I can't believe you people still use cheques, for fuck's sake. And that even with a 6 figure income you're looking at 6-12 months of "credit building" before you can qualify for a cheap car loan without being raped on interest rates.
THAT is patent trolling, and that is not what MS does. Patent trolls don't even attempt to license their patents, or if they do it is at extortionary rates.
Most significantly, patent trolls rarely - if ever - have an actual product in the market that leverages the patent.
This is something I have been wondering too. Doesn't it just lead to applications crashing more often than them normally reporting they cannot allocate more memory?
It results in (practically speaking) non-deterministic behaviour. Which is pretty much the worst thing you could have when it comes to system reliability. The OOM Killer (a solution to a problem that shouldn't even exist) basically kills stuff at random and (at least in my experience) rarely the process that's actually causing the problem in the first place. It's quite common for the OOM Killer to reduce a system to a state where it's impossible to log in at all.
The (default) memory allocation policy in Linux is insane and was only implemented to pander to badly written programs. Fortunately you can make it behave in a sane manner by fiddling a sysctl, but that can (and does) break the aforementioned badly written code (a common victim is the JVM).
(It also allows me to easily amaze and astonish newbies by showing them I'm running 11 GUI programs simultaneously. And don't get me started on the miracle of multiple windows and tabs in browsers...)
11 Windows ? Multiple tabs ? WOW !
Oh, wait, holdon. That wouldn't have amazed me nearly a decade and a half ago (though the tabbed browser would have been a bit novel). Why should it today ?
He's just asserting that it isn't a good sign that their software seems to be growing unnecessarily into a bloated pig. And others will agree with him. 'Fatter' software is by definition less reliable. It involves pushing more electrons around to do the same task as before, and the laws of probability decree that this means more potential points-of-failure in the physical realm.
Your flawed assumption is that the "fatter" software is, in fact, doing "the same task as before".
Maybe I shouldn't respond because you are obviously clueless but anyway. Most consumer routers run Linux, Unix or Maybe Qnx so the percentage is way higher.
You're equating appliances to servers, and *I'm* the one that's clueless ?
Without elevated privileges all the virus is going to do is mess up the user account it was installed in. And that should be backed up.
But typically isn't. To say nothing of an unelevated account being able to do pretty much anything a piece of malware might want to - send spam, host a warez site, participate in a botnet, etc, etc - and that's assuming it can't find a way to elevate itself, either via an exploit or simple social engineering.
Even with de-duplication you need to strip the VM images if you want to maximize savings; the more software you have the more rapidly and thoroughly you'll get desynchronization on patch levels and such.
Large amounts of data in VM images are always going to be the same in a typical environment - essentially all the OS files. They might get out of sync for brief periods if one machine is updated before another, but they'll come back into sync once all machines are brought to the same baseline (which, outside of extraordinary circumstances, they should be).
That is an interesting consideration, yes. Have you seen any thorough review of such benefits in various situations (gains vs. simply adding cache memory in other parts of the chain, etc)?
I seem to recall reading a NetApp paper investigating the benefits. The biggest ones come from better cache usage (thus reducing physical disk I/O) and replication (since deduped blocks only need to be replicated once).
Yep. Many enterprise storage architectures I've seen aren't exactly optimized for customer cost, but rather for maximizing the storage vendors sales revenue.
If you need the features, you have to pay the price. It's the way of the world.
I'm sure it makes sense when, as it often is, you have the worst-case scenario in storage, but I'm not sure it still makes sense if there's even an effort at improving the baseline instead.
The advantage of dedup is it's automatic and constant. "Improving the baseline" requires people-time investment both at the beginning, and as ongoing maintenance, to say nothing of the inconvenience of having to work with "stripped" installations. I'm far more interested in my admins doing productive work than I am them trying to shave a few dozen MBs out of an OS install.
...you also said it was "pretty typical of Windows releases stretching back to, well, basically forever."
I said it was pretty typical for a contemporary Windows release to run acceptably well on a high-end PC dating to ~6-7 years old, and that this has been true basically forever. Which it has:
Windows 3.1 (1992) - needed at least a 386, they were introduced 1985.
Windows 95 (1995) - basically useful config is a 386 with 4-8MB RAM, these would have been available from about 1987.
Windows NT 4 (1996) - basically useful config is a 486 with 16MB RAM, and these first came out in 1989.
Windows 2000 (2000) - basically useful config is a Pentium with 64MB, these first appeared around 1993.
Etc.
And forever means forever.
"Basically forever" is a colloquial phrase. Obviously it's not meant to apply to time periods before Windows or the CPUs to run it even existed.
Typical of Windows releases forever? AHAH! The last Windows PC I bought I got in 2000. It came with a 766MHz Pentium, 128MB RAM, and the graphics was built into the motherboard.
So a middle to low-end PC then. Ie: not relevant to my example.
Do not tell me your configuration was typical.
I didn't even _imply_ it. In fact, I said the complete opposite.
In 2006 I bought a PC with Linux preinstalled. The CPU was a Celeron D, the graphics was built in, and it came with 128 MB RAM and a 40GB HDD.
Bottom of the barrel then. Again, not a relevant example.
While my PCs were budget models mid range desktops didn't come with much more standard, a faster CPU, more RAM, larger HDD, and a dedicated graphics card.
Things that make all the difference in the world. It doesn't take much for a PC to go from "fast" to "slow" - even five years ago, the extra cost would only have been a few hundred $ at most (basically, RAM and video).
However, mid-range desktops aren't relevant to my point. I explicitly said a *high-end* PC dating to 6-7 years old.
Of the various plugs and sockets I've spent time living with (Australian, US, European, British), my personal favourite is the Swiss one. Small, secure, strong and aesthetically pleasing. The habit the Swiss have of also integrating a socket with most light switches is also quite useful.
Was the PC top of the line and maxed out when it came out?
For a home PC, pretty close to it - 933Mhz CPU, 1GB RAM, GeForce 256.
Which is pretty typical of Windows releases stretching back to, well, basically forever. If you have a high-end machine from 6-7 years ago, you'll be able to get acceptable performance out of the contemporary OS release, with maybe some minor upgrades (eg: RAM, video card).
Not that I can recall. This *was* several years ago, though, so I might be mistaken.
It was pretty uninteresting hardware though (Intel CPU + chipset, Intel NIC, etc) so I'd be surprised if it didn't work fine without any third party drivers.
The argument say that the same should be true of programs - instead of trying to keep an up-to-the-second list of all 5 trillion viruses in the world, why not keep a list of the 50 programs that SHOULD be allowed to run, and assume that anything else is bad?
Because it is impossible for the OS vendor to do this effectively, and as soon as the user can do it, the potential security gains disappear (the "dancing bunnies" problem).
This makes logical sense to me, but (apparently) it isn't done. I assume it's much harder than it sounds. Can anyone explain this?
Technically, it's trivial to implement - but the problems with viruses (largely) aren't technical, they're social.
Umm, Linux is the most popular OS for web servers so I would dare say it's popular enough.
That accounts for what, maybe 0.001% of internet-connected machines ?
The issue is that viruses on Linux, Unix and OS X are less destructive because they can only effect the individual user account unless they are able to first infect the user account and then escalate their priviledges to root.
This is, at best, insignificant semantics. What, exactly, do you think the average piece of malicious code needs elevated privileges for ?
Slashdot Mirror
If you are going to do virtualization, the only benefit comes when you invest in a cluster otherwise don't do it at all.
This is not true at all. Indeed, the benefits of virtualisation are such that even for a single service on a single server, it's generally better to make it a VM.
If you decide to go that route regardless, invest in at least 2 dual-port 4GB Fiber HBAs per host server. You'll be thankful in the long run.
Do you have any idea how much disk you need on the back end to exceed the throughput of even a single-port 4Gb HBA, for more than a second or two ?
There are a lot of places to spend money before worrying about how fat your server-side SAN connections are. A *LOT*. 99% of the time, a single 1Gb iSCSI port is more than enough from a performance perspective.
If your take-home pay is >$42,000 a month (very conservative estimate for 1M/year gross), do you really need a car loan?
A "six figure income" starts at $100,000, not $1,000,000.
And if you're in that situation, and you do need a loan, that doesn't speak well for your money management skills.
With low financing rates like 0% - 2% being quite common, I'd be mad *not* to take advantage of it and use the cash somewhere it would get a better return.
We don't use cheques, we use checks. I don't particularly care that you need to re-enforce the differences our languages have developed over the years, most of us are aware that we spell somethings different just so its clear we are NOT YOU. Get over yourself.
You're the only person here who feels the need to "re-enforce the differences". I write it that way because that's the way I've been writing it for thirty years.
I bought my first car with a low interest rate and the car cost more than half my yearly salary, it was the first thing I bought on credit. Not really sure what your problem is, but until about a year ago having no credit wasn't really an issue for any one unless you were old enough that no credit meant you had probably had bad credit and waited for the bad stuff to disappear off your record.
The problem is when you arrive in the USA, you have no credit. The ridiculous part is where this is inescapably equated with having bad credit.
I can't believe you people still haven't gotten over yourselves. Keep it up, at least the rest of the world can see where we got our arrogance from that way. By the way, hows' that world power thing going for you now days?
Apparently you think I'm British. I'm not.
With a "6 figure" income you need to borrow money to buy basic transportation?
No, I just know that my money can be put to better use than buying a car with cash or sitting in deposit for a secured loan. Particularly in the very expensive first few months of an international move (and especially before I receive my first paycheque).
Can I have a $30,000 loan? I have a six-figure income, I can even have you meet my employer to prove it. What's that? You don't like giving loans to people you just meet, even if they make $120k a year?
Yet in many countries (Australia, Switzerland, England, to name the ones I am personally familiar with) you can do just that - simple proof of employment and income is all that's necessary (unless you're self-employed). I had a credit card with a $20k limit within a couple of months of arriving in Switzerland, and the only reason it took that long was the wait for my residency permit to be processed. All they wanted was evidence that I was a legal resident and proof of income. Further, stupidities like a "credit check" with too large an outstanding amount on a card resulting in a loss of "credit rating" were not even a passing concern.
Really, the biggest problem dealing with US banks as a new resident is that anything not American may as well not exist. Five year history with same company ? Tens of thousands of dollars in cash back home ? Existing mortgage with a decade of regular repayments and rental income ? Several existing credit cards ? If you're new to the country it all means diddly squat. You're in the same boat as some random kid just finishing high school who's never had a job - you have to spend 6 months (if you're lucky) to 12 months (more commonly) walking on eggshells. It's ludicrous.
And we didn't fall for that chip-and-PIN-and-you're-liable-if-your-card-is-lost scam.
What "scam" ? I'm not aware of any countries where the customer is liable for fraudulent transactions unless they've been grossly negligent (and being grossly negligent with your "checks" won't produce any different result).
Anyhow, taking a loan to buy a car is almost always a mistake. Save your money, pay cash, be free.
There are numerous situations where it makes more sense to get a loan. Everything from not having sufficient cash on hand, to being able to get a better return on the money elsewhere.
Citation please. I was able to qualify for a low (0%) loan on a new vehicle, with only your normal everyday 5 figure income. Dodge Diesel truck--not some econobox.
Without a credit rating ?
Don't ask me why this is the case...
Because - as I have recently found after starting a job here - the American banking system is utterly insane.
I can't believe you people still use cheques, for fuck's sake. And that even with a 6 figure income you're looking at 6-12 months of "credit building" before you can qualify for a cheap car loan without being raped on interest rates.
THAT is patent trolling, and that is not what MS does. Patent trolls don't even attempt to license their patents, or if they do it is at extortionary rates.
Most significantly, patent trolls rarely - if ever - have an actual product in the market that leverages the patent.
This is something I have been wondering too. Doesn't it just lead to applications crashing more often than them normally reporting they cannot allocate more memory?
It results in (practically speaking) non-deterministic behaviour. Which is pretty much the worst thing you could have when it comes to system reliability. The OOM Killer (a solution to a problem that shouldn't even exist) basically kills stuff at random and (at least in my experience) rarely the process that's actually causing the problem in the first place. It's quite common for the OOM Killer to reduce a system to a state where it's impossible to log in at all.
The (default) memory allocation policy in Linux is insane and was only implemented to pander to badly written programs. Fortunately you can make it behave in a sane manner by fiddling a sysctl, but that can (and does) break the aforementioned badly written code (a common victim is the JVM).
(tolerance is 3% here in Victoria, Australia);
Though note that the *Australian* vehicle design regulations say your speedo only has to be accurate to within 10%.
(It also allows me to easily amaze and astonish newbies by showing them I'm running 11 GUI programs simultaneously. And don't get me started on the miracle of multiple windows and tabs in browsers...)
11 Windows ? Multiple tabs ? WOW !
Oh, wait, holdon. That wouldn't have amazed me nearly a decade and a half ago (though the tabbed browser would have been a bit novel). Why should it today ?
He's just asserting that it isn't a good sign that their software seems to be growing unnecessarily into a bloated pig. And others will agree with him. 'Fatter' software is by definition less reliable. It involves pushing more electrons around to do the same task as before, and the laws of probability decree that this means more potential points-of-failure in the physical realm.
Your flawed assumption is that the "fatter" software is, in fact, doing "the same task as before".
Maybe I shouldn't respond because you are obviously clueless but anyway. Most consumer routers run Linux, Unix or Maybe Qnx so the percentage is way higher.
You're equating appliances to servers, and *I'm* the one that's clueless ?
Without elevated privileges all the virus is going to do is mess up the user account it was installed in. And that should be backed up.
But typically isn't. To say nothing of an unelevated account being able to do pretty much anything a piece of malware might want to - send spam, host a warez site, participate in a botnet, etc, etc - and that's assuming it can't find a way to elevate itself, either via an exploit or simple social engineering.
Even with de-duplication you need to strip the VM images if you want to maximize savings; the more software you have the more rapidly and thoroughly you'll get desynchronization on patch levels and such.
Large amounts of data in VM images are always going to be the same in a typical environment - essentially all the OS files. They might get out of sync for brief periods if one machine is updated before another, but they'll come back into sync once all machines are brought to the same baseline (which, outside of extraordinary circumstances, they should be).
That is an interesting consideration, yes. Have you seen any thorough review of such benefits in various situations (gains vs. simply adding cache memory in other parts of the chain, etc)?
I seem to recall reading a NetApp paper investigating the benefits. The biggest ones come from better cache usage (thus reducing physical disk I/O) and replication (since deduped blocks only need to be replicated once).
Yep. Many enterprise storage architectures I've seen aren't exactly optimized for customer cost, but rather for maximizing the storage vendors sales revenue.
If you need the features, you have to pay the price. It's the way of the world.
I'm sure it makes sense when, as it often is, you have the worst-case scenario in storage, but I'm not sure it still makes sense if there's even an effort at improving the baseline instead.
The advantage of dedup is it's automatic and constant. "Improving the baseline" requires people-time investment both at the beginning, and as ongoing maintenance, to say nothing of the inconvenience of having to work with "stripped" installations. I'm far more interested in my admins doing productive work than I am them trying to shave a few dozen MBs out of an OS install.
www.google.com microsoft and viruses and "thing of the past".
There is nothing on the first page of results to support the claim.
I said it was pretty typical for a contemporary Windows release to run acceptably well on a high-end PC dating to ~6-7 years old, and that this has been true basically forever. Which it has:
Windows 3.1 (1992) - needed at least a 386, they were introduced 1985.
Windows 95 (1995) - basically useful config is a 386 with 4-8MB RAM, these would have been available from about 1987.
Windows NT 4 (1996) - basically useful config is a 486 with 16MB RAM, and these first came out in 1989.
Windows 2000 (2000) - basically useful config is a Pentium with 64MB, these first appeared around 1993.
Etc.
And forever means forever.
"Basically forever" is a colloquial phrase. Obviously it's not meant to apply to time periods before Windows or the CPUs to run it even existed.
Goddamn you zealots are hard work to deal with.
Typical of Windows releases forever? AHAH! The last Windows PC I bought I got in 2000. It came with a 766MHz Pentium, 128MB RAM, and the graphics was built into the motherboard.
So a middle to low-end PC then. Ie: not relevant to my example.
Do not tell me your configuration was typical.
I didn't even _imply_ it. In fact, I said the complete opposite.
In 2006 I bought a PC with Linux preinstalled. The CPU was a Celeron D, the graphics was built in, and it came with 128 MB RAM and a 40GB HDD.
Bottom of the barrel then. Again, not a relevant example.
While my PCs were budget models mid range desktops didn't come with much more standard, a faster CPU, more RAM, larger HDD, and a dedicated graphics card.
Things that make all the difference in the world. It doesn't take much for a PC to go from "fast" to "slow" - even five years ago, the extra cost would only have been a few hundred $ at most (basically, RAM and video).
However, mid-range desktops aren't relevant to my point. I explicitly said a *high-end* PC dating to 6-7 years old.
Of the various plugs and sockets I've spent time living with (Australian, US, European, British), my personal favourite is the Swiss one. Small, secure, strong and aesthetically pleasing. The habit the Swiss have of also integrating a socket with most light switches is also quite useful.
Was the PC top of the line and maxed out when it came out?
For a home PC, pretty close to it - 933Mhz CPU, 1GB RAM, GeForce 256.
Which is pretty typical of Windows releases stretching back to, well, basically forever. If you have a high-end machine from 6-7 years ago, you'll be able to get acceptable performance out of the contemporary OS release, with maybe some minor upgrades (eg: RAM, video card).
With no third party software/drivers/etc?
Not that I can recall. This *was* several years ago, though, so I might be mistaken.
It was pretty uninteresting hardware though (Intel CPU + chipset, Intel NIC, etc) so I'd be surprised if it didn't work fine without any third party drivers.
The argument say that the same should be true of programs - instead of trying to keep an up-to-the-second list of all 5 trillion viruses in the world, why not keep a list of the 50 programs that SHOULD be allowed to run, and assume that anything else is bad?
Because it is impossible for the OS vendor to do this effectively, and as soon as the user can do it, the potential security gains disappear (the "dancing bunnies" problem).
This makes logical sense to me, but (apparently) it isn't done. I assume it's much harder than it sounds. Can anyone explain this?
Technically, it's trivial to implement - but the problems with viruses (largely) aren't technical, they're social.
Not really. First, the most it could do is infect your own files, not the system.
So only the most important files on the system, then ?
Second, you would have to run it - it can't spread by itself.
Just like most Windows "viruses", you mean ?
Do people running linux run strange executable binaries that people send them?
If most people running Linux were like most people running Windows, they would.
No. It's not like Windows, where reading your email can infect your machine.
No, it's more like opening a PDF could infect your machine.
Umm, Linux is the most popular OS for web servers so I would dare say it's popular enough.
That accounts for what, maybe 0.001% of internet-connected machines ?
The issue is that viruses on Linux, Unix and OS X are less destructive because they can only effect the individual user account unless they are able to first infect the user account and then escalate their priviledges to root.
This is, at best, insignificant semantics. What, exactly, do you think the average piece of malicious code needs elevated privileges for ?