First off, piracy cannot be stopped. Not by technological means, and not by legal means. Some people simply refuse to pay for it, and they will always pirate. There is nothing you can do to stop it, and you already have all the legal means necessary to address that issue.
You can limit piracy to a small enough percentage that it doesn't materially affect sales. Make it easy to buy, easy to use, and cheap enough that, for most people, it's not worth the risk of getting caught making illegal copies.
That's the only thing that has ever kept piracy under control, regardless of technology. Printed books were cheaper and more convenient than hand copied books. When the photocopier came out, it was generally more expensive (and tedious) to copy a book than to just buy a legitimate copy. Records and tapes were cheap, piracy wasn't a major issue. And while CDs were more expensive than records and tapes, they offered greater quality, greater durability, and no easy way to copy them while maintaining the quality, so piracy in CDs was mostly from professional counterfeiting groups (whom you have the legal tools to stop). There was no DRM, you could make personal use cassettes and MP3s from your CDs.
Piracy started growing in the VCR age, because the movies were expensive. So, they introduced MacroVision, and the copy-prevention arms race began. They continued it with DVDs using CSS, and high release prices. Professional counterfeiting soared. Repeat mistake with HD-DVD and Blu-ray Disc. Broadband internet access became common, and with prices still high for DVDs, digital piracy of DVDs started growing. People wanted to watch their DVD based movies on their newer portable digital devices, but they had no way to transfer the content, then they found they could download those movies. And if they were going to have to download a copy anyway, why buy the overpriced original that they weren't going to use?
Started to repeat the same mistake with digital music downloads, but eventually conceded on DRM. Notice what happened to sales after DRM was dropped from some labels starting in Jan 2007, they doubled, then doubled again in 2008, then when all the labels agreed to DRM free iTunes+ downloads in 2009, sales doubled again. How many billions of songs is Apple legally selling every year? ~4B. Granted, not all of that increase was due solely to removing DRM, but that was a key part of it. Apple's iTunes Store has also sold millions of feature length movies and hundreds of millions of TV episodes. Then, there is Amazon, licensed streaming music services, and other sellers.
TL;DR Make it convenient, DRM free, and reasonably priced and 95%-99% of the potential market will pay for it. The ~1% who are committed to piracy will copy it no matter what you do. Technology changes rapidly, people are not willing to pay for the same material in a new format every few years, unless it's very cheap to do so. Until content distributors adapt a sales model that allows people to use their licensed media with any device they own that is capable of playing it, as many times as they wish to play it (or have a reasonable pay-per-view/rental model), piracy will continue to grow. All the attempts to limit it using DRM, technology, or laws will fail to slow piracy, in fact, they increase the incentive to seek out DRM free versions that are usually only available via "piracy". Resist that, and you'll soon find the market has gone elsewhere.
Agreed. However, as we're all well aware, most users completely ignore the permissions Android apps (and Facebook Apps, etc.) request when installing. That's only useful info to an awake, aware, knowledgeable user. All apps should first be vetted by someone qualified to determine if the requested permissions or behaviors represent an inappropriate level of risk (Apple does some of that, but not enough), before they are placed on a store such as Google Play. Apple's approach is better for 95+% of users, but it's not best for everyone.
This is why republics are superior to democracies. Democracy is tyranny of the people. A republic is governed by rule of law, a democracy is governed by the fickle will of the masses.
The argument exists solely because the proponents of these vehicles talk about how cheap they are to drive. It's a lie. If they would stop lying about it, then companies like CR & Edmunds would do their analysis, and 95% of the population would ignore those and simply buy what they want.
The problem with the Volt is that it's priced about $10k too high. The only reason anyone will ever get near breakeven is because of the tax credit. Take away that $7500 (which expires in the quarter that they sell the 60,000'th vehicle, and is lower the next quarter), eventually dropping to zero. The problem is that the tax credit isn't "free", it's taking money from every taxpayer, including the person buying the car. 60k cars & $7500 = $450M, plus more for the decreasing credits. It's bad energy policy, and bad tax policy.
A Prius is $22k. The also compare it to the all electric Leaf, and the all gas Chevy Cruze.
The battery probably will last longer than the warranty period, but did you read where Chevy estimates that it's capacity will be reduced 10%-30% in those 100k miles? Oh, BTW, that's 100k vehicle miles, not 100k electric miles. By the time you hit 100k-150k electric miles, the battery is expected to need replacement, if for no other reason than significant loss of capacity.
Apple on the other hand, is not allowing users to configure IPv6 - even if it is present and enabled, what good does it do if it can't be configured?
Wrong. Fundamentally wrong. As I explained in my first post, the first release of this new version of the utility doesn't allow you to configure it, but the old version does, and it still works. It's a non-issue for 99% of users because most people can't yet get IPv6 support from their ISP. The device still fully supports IPv6, but if you need it right now, you have to use the older version of the configuration utility, that's all.
Hold on a minute, you're not done yet. You didn't count charging losses or taxes on the electricity yet. And the Volt gets about 3.4mi/KWh, so your figures for the Volt costs start off 15% low. Consumer Reports calculates the Volt costs $0.038/mi @ $0.11/KWh, 12/11th of that is $0.0415, so let's use $0.04/mi for simplicity. Ok, savings is now $0.12/mi, so were now at 83k mi for BE on $10k difference, 150k mi on $18k difference. That's counting exclusively electric miles on the Volt, gasoline powered miles don't count, because they cost about the same as the other vehicles.
What if we reduce the gasoline cost to $6? That's $0.12/mi - $0.04/mi = $0.08/mi difference. Now, we're at 125k mi for $10k, and $ 225k mi for $18k.
But wait, the Volt battery pack wears out. It's guaranteed for 8yrs or 100k mi, but after that, you're looking at replacement when you get near breakeven, even with $8/gal gas, and you're not likely to make it to break even with $6/gal gas.
If gasoline remains below ~ $7/gal, or electricity increases, well, you'll never hit breakeven. Edmunds, Consumer Reports, and others have analyzed it extensively, see the first link above for more details.
Multiple "gods", sort of polytheistic IT. If they're good, they'll notice If one of the others isn't doing his/her job, and they'll notice artifacts if one of them is trying to cover his/her tracks.
...do yourself and your admin a favor and get rid of him/her. He/she won't like working for someone who doesn't trust him/her, and you won't like constantly being suspicious.
I've given that advice to all my clients over the years. You can extend the concept to the rest of your IT and/or security team. That doesn't mean you shouldn't take precautions, have checks and balances in place, etc, but fundamentally, if there isn't a high level of trust, deal with the lack of trust, either by discussing it until there is an understanding and trust, or by ending the relationship.
Secretive monitoring is not the way to handle a lack of trust. The only exception is when there is already probable cause to believe a crime has been committed, then, in some cases, monitoring to gather proof may or may not be necessary or appropriate.
MS seen as backpedaling on it's support for 64-bit computing over Windows 8 only supporting 32-bit CPUs in tablets.
Come on people, this isn't backpedaling, it's a completely new version of a utility that in it's initial release supports what's in use in 99% of installations. Those who are actually using IPv6 can use the older version until this one adds support (probably in the next release).
5 of those 6 apps listed give you a warning and/or choice before they touch your contacts. Path is the only one that does it without your consent.
I only have one of those 6 installed (FB), and I did not give it permission to access and synchronize my contacts, and I never will.
As others pointed out, comparing that to malware is more than a stretch. You could make a case for Path qualifying because it did so without notification or consent. At most, that's one app that qualifies. Even if you do count it as malware, comparing it to malware that sends SMS messages that cost you money is absurd.
If you want to point out malware on iOS, you should point to the 2-3 actual cases of malware that have been found in the App Store over the years, not 5 applications that notify you they're going to access your contacts.
No, bottleneck wasn't the bandwidth to the server. Single machine connected to it's switch via 100mb, no other users (after hours), performance for one app more than doubled when the server to switch and sw-sw links were upgraded to 1gb. Latency was this issue, not bandwidth. Bandwidth to the user was unchanged, and with no other users, it wasn't a sharing issue.
There were certain reports where shared bandwidth for multiple users was a constraining factor, and in those cases, the extra bandwidth of the server and switch links helped as well during normal work hours since no user could saturate switch to switch or switch to server links.
For small packets the latency difference between 100 Mb and 1000 Mb is negligible until you start pushing against your bandwidth caps.
Incorrect. Using small packets (256 byte data payload, plus packet overhead), performance for a latency limited application improved 35% to 75% (depending upon the speed of the workstation CPU) by upgrading the sw-sw and sw-server links to 1gb, even with the WS remaining on 100mb. Again, all tests performed after hours with only a single user using the system.
Example: (assumes store-and-forward switches, cut-through switching is rare because it's impractical when switching to a faster link. Typical latency in the switch is below 10ns when the outgoing link is available, so it has been ignored) With 256byte packets, in a latency limited (e.g. response to each packet is required), it's impossible to approach the bandwidth of any link, you can't exceed 50% of bandwidth in theory, and in practice quite a bit less. To put real numbers to it, a 256byte (data) packet plus overhead will be in the vicinity of 280bytes. That's 2240 bits, round it to 2250 bit times for simplicity. @ 100mb/s, that's 22.5us. @ 1gb/s, it's 2.25us. Add 0.25us for propagation delay on an ~ 50m cable (@.66c), so those become 24.75us and 2.5us Now, calculate round trip latencies for (not counting server response time or ~ 100): a. ws-(100mb)-sw-(100mb)-sw-(100mb)-svr = 24.75+24.75+24.75us each way, 148.5us round trip b. ws-(100mb)-sw-(1gb)-sw-(1gb)-svr = 24.75+2.5+2.5us each way, 59.5us round trip.
That's significant. In fact, it's so significant, that by timing round trip delay between a known "client" and a known "server", I can tell with a high level of confidence how many 10mb, 100mb, and 1gb (or faster) links are between the two, and what are the speed of those links (but not which link is which speed). It's not precise enough to reliably identify links above 1gb.
As packet sizes shrink to the minimum 64byte Ethernet packet the total time get's smaller, but the ratios don't change much. How much difference in measured performance depends upon the ws and server processing latencies, but as my tests show, it can make a significant real world difference even with relatively slow (P3/450) machines as both client and server. Faster machines will have lower ws & server processing times, making those network latencies more significant. Clearly, taking 1gb to the ws lowers latency further, and my tests show some performance boost for doing so, but you start be become limited by the processing latencies of the ws and/or server so the gains are less dramatic. The improvement doing to 10gb/s is much less dramatic.
I have seen cases where places had daisy-chained switches and hubs for network capacity. This can become a real latency issue as every switch take a little time to process each packet.
The internal latency in most switches is trivial, performance is very similar connecting server and ws to the same switch vs having 2 or 3 switches between the server and ws, IF and only IF the sw-sw links are faster than the ws-switch link. The limiting factor is the time to buffer and retransmit packet on the uplink. If the sw-sw links are the same speed as the ws-switch link, additional hops can have a notable effect, especially at link rates below 1gb/s. Once you reach 1gb/
I agree. They also don't mention that the pay at startups isn't always great. The potential payout can be great, but the actual pay is often on the lower end of the scale.
Still, I love working for startups and small companies, so much less BS to deal with.
Yes, with the slowing population growth and the increasing average life span, it is heading for a time when it won't bring in more than it costs. Under it's current rules, it's not sustainable.
And, while I'm in favor of a balanced budget, a BBA needs to allow some flexibility. In times of war, emergency, or recession, gov't spending may need to do exceed revenues. The problem is that we've had almost continual deficits for 40 years. An occasional deficit for 5 years can be managed, but 40 years can't.
I stand corrected, but that doesn't materially change the situation. SS still brings in more revenue than it costs, and Military and "discretionary" spending are each near 20% of the budget. Cutting out the "wars" would make a huge difference.
The problem with the space program is US funding. Increasingly citizens in the US want their entitlements without any strings attached.
Despite all the debates and rhetoric about it, entitlements aren't the problem. Social Security and Medicare the two big entitlements are in fact paid for from separate taxes that exceed the amount spent on those programs. Look at the federal budget. Military spending, is the biggest portion, bigger than all entitlements combined. NASA's budget is less than 1% of the federal budget. What's killing us are all the "wars", the overseas wars, the "war on drugs", the "war on terror", etc.
Don't misunderstand me, we need a military, we need defense. But the "war on drugs" is a complete waste, the "war or terror" is out of control, and the other wars are just a way for people supplying the military to get rich while bankrupting the country.
Mars Pathfinder demonstrated a model that works well. If you're going with the "Launch Early, Launch Often" model I suggested, then you don't have to try to cram the most advanced (and most expensive) technologies into each mission, you go with lower cost and simpler instruments. One difference from Pathfinder, these instruments have to be designed for longevity, but they don't have to survive the heat and g-forces of a landing, so you have different engineering tradeoffs.
Again, the frequent launch strategy also reduces the scientific loss from any failure, whether it's a launch failure, or an early instrument failure, etc. As long as we can communicate with it and at least one instrument is working, we're getting potentially useful info. It also means that the complete loss of a probe doesn't create a multi-decade gap in information gathering (with a corresponding loss of the people skilled in building the craft, operating them, and analyzing the data.
I suspect you can operate such a program for well under $500M/yr, maybe under $250M/yr. Those are "wild guesses" based upon the fact that we were able design, build, launch, and operate Mars Pathfinder for $280M total over several years. Allowing more for operations, more for build costs, and launching 1-2 probes ~ every 5 years, I suspect $250M/yr is on the high side, but since I have no other hard data, I've allowed for twice that on the high end.
Depends upon what OS you were running, and what the server was doing, that might have been fine. A 486/66 w32MB running Netware 3/4 was a pretty good server, didn't need much CPU. Might have benefited from more RAM, depending upon your environment. Of course, the CEO's machine was wasted, but that's true in about 70% of the companies I've seen.
Now, if you were running a Windows server, especially one running SQL Server, Exchange, or Oracle on 32MB RAM, you were hurting for RAM.
Slashdot Mirror
First off, piracy cannot be stopped. Not by technological means, and not by legal means. Some people simply refuse to pay for it, and they will always pirate. There is nothing you can do to stop it, and you already have all the legal means necessary to address that issue.
You can limit piracy to a small enough percentage that it doesn't materially affect sales. Make it easy to buy, easy to use, and cheap enough that, for most people, it's not worth the risk of getting caught making illegal copies.
That's the only thing that has ever kept piracy under control, regardless of technology. Printed books were cheaper and more convenient than hand copied books. When the photocopier came out, it was generally more expensive (and tedious) to copy a book than to just buy a legitimate copy. Records and tapes were cheap, piracy wasn't a major issue. And while CDs were more expensive than records and tapes, they offered greater quality, greater durability, and no easy way to copy them while maintaining the quality, so piracy in CDs was mostly from professional counterfeiting groups (whom you have the legal tools to stop). There was no DRM, you could make personal use cassettes and MP3s from your CDs.
Piracy started growing in the VCR age, because the movies were expensive. So, they introduced MacroVision, and the copy-prevention arms race began. They continued it with DVDs using CSS, and high release prices. Professional counterfeiting soared. Repeat mistake with HD-DVD and Blu-ray Disc. Broadband internet access became common, and with prices still high for DVDs, digital piracy of DVDs started growing. People wanted to watch their DVD based movies on their newer portable digital devices, but they had no way to transfer the content, then they found they could download those movies. And if they were going to have to download a copy anyway, why buy the overpriced original that they weren't going to use?
Started to repeat the same mistake with digital music downloads, but eventually conceded on DRM. Notice what happened to sales after DRM was dropped from some labels starting in Jan 2007, they doubled, then doubled again in 2008, then when all the labels agreed to DRM free iTunes+ downloads in 2009, sales doubled again. How many billions of songs is Apple legally selling every year? ~4B. Granted, not all of that increase was due solely to removing DRM, but that was a key part of it. Apple's iTunes Store has also sold millions of feature length movies and hundreds of millions of TV episodes. Then, there is Amazon, licensed streaming music services, and other sellers.
TL;DR
Make it convenient, DRM free, and reasonably priced and 95%-99% of the potential market will pay for it. The ~1% who are committed to piracy will copy it no matter what you do. Technology changes rapidly, people are not willing to pay for the same material in a new format every few years, unless it's very cheap to do so. Until content distributors adapt a sales model that allows people to use their licensed media with any device they own that is capable of playing it, as many times as they wish to play it (or have a reasonable pay-per-view/rental model), piracy will continue to grow. All the attempts to limit it using DRM, technology, or laws will fail to slow piracy, in fact, they increase the incentive to seek out DRM free versions that are usually only available via "piracy". Resist that, and you'll soon find the market has gone elsewhere.
"Robert Krampf, who runs the web site 'The Happy Scientist,' ...
I read his blog post, Robert doesn't sound so happy.
You should read the definitions of republic and democracy. You might find them informative.
Democratic republics are the best form of government yet tried. But pure democracy is impractical when you have hundreds of thousands of citizens.
Agreed. However, as we're all well aware, most users completely ignore the permissions Android apps (and Facebook Apps, etc.) request when installing. That's only useful info to an awake, aware, knowledgeable user. All apps should first be vetted by someone qualified to determine if the requested permissions or behaviors represent an inappropriate level of risk (Apple does some of that, but not enough), before they are placed on a store such as Google Play. Apple's approach is better for 95+% of users, but it's not best for everyone.
This is why republics are superior to democracies. Democracy is tyranny of the people. A republic is governed by rule of law, a democracy is governed by the fickle will of the masses.
The argument exists solely because the proponents of these vehicles talk about how cheap they are to drive. It's a lie. If they would stop lying about it, then companies like CR & Edmunds would do their analysis, and 95% of the population would ignore those and simply buy what they want.
The problem with the Volt is that it's priced about $10k too high. The only reason anyone will ever get near breakeven is because of the tax credit. Take away that $7500 (which expires in the quarter that they sell the 60,000'th vehicle, and is lower the next quarter), eventually dropping to zero. The problem is that the tax credit isn't "free", it's taking money from every taxpayer, including the person buying the car. 60k cars & $7500 = $450M, plus more for the decreasing credits. It's bad energy policy, and bad tax policy.
A Prius is $22k. The also compare it to the all electric Leaf, and the all gas Chevy Cruze.
The battery probably will last longer than the warranty period, but did you read where Chevy estimates that it's capacity will be reduced 10%-30% in those 100k miles? Oh, BTW, that's 100k vehicle miles, not 100k electric miles. By the time you hit 100k-150k electric miles, the battery is expected to need replacement, if for no other reason than significant loss of capacity.
Apple on the other hand, is not allowing users to configure IPv6 - even if it is present and enabled, what good does it do if it can't be configured?
Wrong. Fundamentally wrong. As I explained in my first post, the first release of this new version of the utility doesn't allow you to configure it, but the old version does, and it still works. It's a non-issue for 99% of users because most people can't yet get IPv6 support from their ISP. The device still fully supports IPv6, but if you need it right now, you have to use the older version of the configuration utility, that's all.
Hold on a minute, you're not done yet. You didn't count charging losses or taxes on the electricity yet. And the Volt gets about 3.4mi/KWh, so your figures for the Volt costs start off 15% low. Consumer Reports calculates the Volt costs $0.038/mi @ $0.11/KWh, 12/11th of that is $0.0415, so let's use $0.04/mi for simplicity. Ok, savings is now $0.12/mi, so were now at 83k mi for BE on $10k difference, 150k mi on $18k difference. That's counting exclusively electric miles on the Volt, gasoline powered miles don't count, because they cost about the same as the other vehicles.
What if we reduce the gasoline cost to $6? That's $0.12/mi - $0.04/mi = $0.08/mi difference. Now, we're at 125k mi for $10k, and $ 225k mi for $18k.
But wait, the Volt battery pack wears out. It's guaranteed for 8yrs or 100k mi, but after that, you're looking at replacement when you get near breakeven, even with $8/gal gas, and you're not likely to make it to break even with $6/gal gas.
If gasoline remains below ~ $7/gal, or electricity increases, well, you'll never hit breakeven. Edmunds, Consumer Reports, and others have analyzed it extensively, see the first link above for more details.
Multiple "gods", sort of polytheistic IT. If they're good, they'll notice If one of the others isn't doing his/her job, and they'll notice artifacts if one of them is trying to cover his/her tracks.
...do yourself and your admin a favor and get rid of him/her. He/she won't like working for someone who doesn't trust him/her, and you won't like constantly being suspicious.
I've given that advice to all my clients over the years. You can extend the concept to the rest of your IT and/or security team. That doesn't mean you shouldn't take precautions, have checks and balances in place, etc, but fundamentally, if there isn't a high level of trust, deal with the lack of trust, either by discussing it until there is an understanding and trust, or by ending the relationship.
Secretive monitoring is not the way to handle a lack of trust. The only exception is when there is already probable cause to believe a crime has been committed, then, in some cases, monitoring to gather proof may or may not be necessary or appropriate.
It's likely to be less costly than an appeal, and they can finally put it behind them.
Woosh!!!
MS seen as backpedaling on it's support for 64-bit computing over Windows 8 only supporting 32-bit CPUs in tablets.
Come on people, this isn't backpedaling, it's a completely new version of a utility that in it's initial release supports what's in use in 99% of installations. Those who are actually using IPv6 can use the older version until this one adds support (probably in the next release).
It's still experimental, you may be able to get in as a study patient.
And 2/3 of the people in my office are orthodox Jews. If you just extrapolate that, there are 200M Orthodox Jews in the USA.
In other words, the sample size is too small and too selective to be of any use and your comment is complete nonsense.
At least not in the US, but I'm pretty sure that applies to Canada also. If they did indeed crowd-source the data, this lawsuit should be DOA.
5 of those 6 apps listed give you a warning and/or choice before they touch your contacts. Path is the only one that does it without your consent.
I only have one of those 6 installed (FB), and I did not give it permission to access and synchronize my contacts, and I never will.
As others pointed out, comparing that to malware is more than a stretch. You could make a case for Path qualifying because it did so without notification or consent. At most, that's one app that qualifies. Even if you do count it as malware, comparing it to malware that sends SMS messages that cost you money is absurd.
If you want to point out malware on iOS, you should point to the 2-3 actual cases of malware that have been found in the App Store over the years, not 5 applications that notify you they're going to access your contacts.
No, bottleneck wasn't the bandwidth to the server. Single machine connected to it's switch via 100mb, no other users (after hours), performance for one app more than doubled when the server to switch and sw-sw links were upgraded to 1gb. Latency was this issue, not bandwidth. Bandwidth to the user was unchanged, and with no other users, it wasn't a sharing issue.
There were certain reports where shared bandwidth for multiple users was a constraining factor, and in those cases, the extra bandwidth of the server and switch links helped as well during normal work hours since no user could saturate switch to switch or switch to server links.
For small packets the latency difference between 100 Mb and 1000 Mb is negligible until you start pushing against your bandwidth caps.
Incorrect. Using small packets (256 byte data payload, plus packet overhead), performance for a latency limited application improved 35% to 75% (depending upon the speed of the workstation CPU) by upgrading the sw-sw and sw-server links to 1gb, even with the WS remaining on 100mb. Again, all tests performed after hours with only a single user using the system.
Example: (assumes store-and-forward switches, cut-through switching is rare because it's impractical when switching to a faster link. Typical latency in the switch is below 10ns when the outgoing link is available, so it has been ignored)
With 256byte packets, in a latency limited (e.g. response to each packet is required), it's impossible to approach the bandwidth of any link, you can't exceed 50% of bandwidth in theory, and in practice quite a bit less. To put real numbers to it, a 256byte (data) packet plus overhead will be in the vicinity of 280bytes. That's 2240 bits, round it to 2250 bit times for simplicity. @ 100mb/s, that's 22.5us. @ 1gb/s, it's 2.25us. Add 0.25us for propagation delay on an ~ 50m cable (@.66c), so those become 24.75us and 2.5us Now, calculate round trip latencies for (not counting server response time or ~ 100):
a. ws-(100mb)-sw-(100mb)-sw-(100mb)-svr = 24.75+24.75+24.75us each way, 148.5us round trip
b. ws-(100mb)-sw-(1gb)-sw-(1gb)-svr = 24.75+2.5+2.5us each way, 59.5us round trip.
That's significant. In fact, it's so significant, that by timing round trip delay between a known "client" and a known "server", I can tell with a high level of confidence how many 10mb, 100mb, and 1gb (or faster) links are between the two, and what are the speed of those links (but not which link is which speed). It's not precise enough to reliably identify links above 1gb.
As packet sizes shrink to the minimum 64byte Ethernet packet the total time get's smaller, but the ratios don't change much. How much difference in measured performance depends upon the ws and server processing latencies, but as my tests show, it can make a significant real world difference even with relatively slow (P3/450) machines as both client and server. Faster machines will have lower ws & server processing times, making those network latencies more significant. Clearly, taking 1gb to the ws lowers latency further, and my tests show some performance boost for doing so, but you start be become limited by the processing latencies of the ws and/or server so the gains are less dramatic. The improvement doing to 10gb/s is much less dramatic.
I have seen cases where places had daisy-chained switches and hubs for network capacity. This can become a real latency issue as every switch take a little time to process each packet.
The internal latency in most switches is trivial, performance is very similar connecting server and ws to the same switch vs having 2 or 3 switches between the server and ws, IF and only IF the sw-sw links are faster than the ws-switch link. The limiting factor is the time to buffer and retransmit packet on the uplink. If the sw-sw links are the same speed as the ws-switch link, additional hops can have a notable effect, especially at link rates below 1gb/s. Once you reach 1gb/
I agree. They also don't mention that the pay at startups isn't always great. The potential payout can be great, but the actual pay is often on the lower end of the scale.
Still, I love working for startups and small companies, so much less BS to deal with.
Yes, with the slowing population growth and the increasing average life span, it is heading for a time when it won't bring in more than it costs. Under it's current rules, it's not sustainable.
And, while I'm in favor of a balanced budget, a BBA needs to allow some flexibility. In times of war, emergency, or recession, gov't spending may need to do exceed revenues. The problem is that we've had almost continual deficits for 40 years. An occasional deficit for 5 years can be managed, but 40 years can't.
I stand corrected, but that doesn't materially change the situation. SS still brings in more revenue than it costs, and Military and "discretionary" spending are each near 20% of the budget. Cutting out the "wars" would make a huge difference.
The problem with the space program is US funding. Increasingly citizens in the US want their entitlements without any strings attached.
Despite all the debates and rhetoric about it, entitlements aren't the problem. Social Security and Medicare the two big entitlements are in fact paid for from separate taxes that exceed the amount spent on those programs. Look at the federal budget. Military spending, is the biggest portion, bigger than all entitlements combined. NASA's budget is less than 1% of the federal budget. What's killing us are all the "wars", the overseas wars, the "war on drugs", the "war on terror", etc.
Don't misunderstand me, we need a military, we need defense. But the "war on drugs" is a complete waste, the "war or terror" is out of control, and the other wars are just a way for people supplying the military to get rich while bankrupting the country.
Mars Pathfinder demonstrated a model that works well. If you're going with the "Launch Early, Launch Often" model I suggested, then you don't have to try to cram the most advanced (and most expensive) technologies into each mission, you go with lower cost and simpler instruments. One difference from Pathfinder, these instruments have to be designed for longevity, but they don't have to survive the heat and g-forces of a landing, so you have different engineering tradeoffs.
Again, the frequent launch strategy also reduces the scientific loss from any failure, whether it's a launch failure, or an early instrument failure, etc. As long as we can communicate with it and at least one instrument is working, we're getting potentially useful info. It also means that the complete loss of a probe doesn't create a multi-decade gap in information gathering (with a corresponding loss of the people skilled in building the craft, operating them, and analyzing the data.
I suspect you can operate such a program for well under $500M/yr, maybe under $250M/yr. Those are "wild guesses" based upon the fact that we were able design, build, launch, and operate Mars Pathfinder for $280M total over several years. Allowing more for operations, more for build costs, and launching 1-2 probes ~ every 5 years, I suspect $250M/yr is on the high side, but since I have no other hard data, I've allowed for twice that on the high end.
Depends upon what OS you were running, and what the server was doing, that might have been fine. A 486/66 w32MB running Netware 3/4 was a pretty good server, didn't need much CPU. Might have benefited from more RAM, depending upon your environment. Of course, the CEO's machine was wasted, but that's true in about 70% of the companies I've seen.
Now, if you were running a Windows server, especially one running SQL Server, Exchange, or Oracle on 32MB RAM, you were hurting for RAM.