I disagree. Progressive taxation simply aligns income tax more with disposable income than gross income. Ideally, there would be a straightforward way to calculate disposable income, but sadly there is not. As a percentage of disposable income, the poor are paying a disproportionately greater amount of tax than the wealthy.
It also stands to reason that wealthier people are receiving greater benefit from societal structures, so it makes sense for them to make greater contributions to support those structures accordingly.
These are tenanted data centers, with customers on long-term contracts. It would take up to 3 years for contracts to expire before they could fully repurpose these facilities.
Also, Verizon/Terremark owns the main interconnection point in Miami, including the Nap of The Americas peering exchange, where the vast majority of traffic to Latin America traverses through. If Microsoft were to simply repurpose that facility for internal use, they'd effectively cut off Latin America from the rest of the Internet.
It's much more likely for a company in this market to be the acquirer. Someone like Equinix, Zayo, Internap, Coresite, or Cologix. 2.5B would be a lot for any one of these companies to shell out. There may also be some realty companies who specialize in leasing space for data centers like Digital Realty Trust who may be interested.
Ultimately though, I think Verizon may need to break up some of these assets if no one company can pony up the entire sum.
One big, big name is Youtube, who was burning through other people's money faster than a drunk Kennedy and getting rightfully sued every 5 minutes for copyright infringement. They had a cool idea, and a completely non-sustainable business model that was guaranteed to put them belly-up within 36 months until Google bought them. Google brought to bear their expertise in funding a free service in a way that keeps customers happy (aka the best targeted advertising available) , allowing YouTube to survive and thrive rather than burning away investors' money until investors got sick of it and'the whole thing imploded.
YouTube's business model from the beginning was to get bought by Google, and so they focused their business towards that end and succeeded. Had that not been their primary goal, they may have done things very differently. In any event, using them as an example of a smaller company that needed to be bought isn't wrong, but the assertion that they needed the good processes that came from the larger company is false. They were very efficient at exiting for maximum return for their shareholders.
The issue is not about ISP's being forced to peer with content providers like Netflix whether they like it or not. It's the major eyeball networks like Comcast and AT&T deliberately avoiding upgrading peering capacity, so that they can strong-arm Netflix into paid peering agreements instead. Let's not forget that Comcast is also trying to pressure Tier 1 transit providers like Level3 to pay for access to their eyeballs now, even though the miles of fiber in the ground that they operate, and thus their share of transport costs, pale in comparison.
As a smaller provider, you should most certainly want to peer with any content network with a significant amount of traffic to your network, as you would have to pay for transit otherwise. There is no way you'd be paying more for cross-connects than you would for transit, and the equipment costs are irrelevant because you're doing that same traffic regardless, whether it comes through a transit or a peer. Your traffic levels aren't going to magically increase when you start peering, unless your existing transit is congested, which is something you should most definitely want to avoid.
Netflix does most of its heavy lifting through its own CDN boxes running FreeBSD + nginx, which it will place on ISPs' networks for free, although the ISP would be responsible for the space and power. They most certainly do have their own data centres though, with infrastructure that goes well beyond Amazon.
I call BS. We're at the Westin as well, and x-connects in the 19th floor meet-me room are free, while x-connects between floors are a one-time build cost. even if you don't have your own space and are leasing lines from a data centre, they're only $50-$200/mo. Powered equipment may not be permitted in the meet-me room, but there is nothing that would stop you from putting some passive DWDM muxes in there so you could even run 40-80x 10Gb-100Gb links over the same fiber. The only way I see x-connects being expensive is you're using Equinix, but that would be your own business decision, and not inherent costs to peering.
Also, 10Gb optics sourced directly from China where they're manufactured can be purchased for less than $100 for LR optics, or less than $1000 for DWDM optics, which is peanuts for a one-time cost. Sure, line cards and therefore physical interfaces are expensive, but that's why public peering exchanges like the SIX (Seattle Internet Exchange) exist so you can peer with many different networks over the same interface. If you are doing enough traffic over a link to justify private peering however, your equipment costs should be built-in to your business model, and whether it's peering or other traffic shouldn't matter.
Although I do agree with your assessment that this isn't a net neutrality issue, it most certainly is a case of large eyeball networks trying to double-dip on charging for bandwidth, and criticism of them is well deserved.
This. Instead of pushing young people towards a certain path, and converting highschools and earlier into trade schools, why not just give them the best, all-rounded education possible and allow them to decide for themselves what they want to do? That's not to say we shouldn't teach them the value of more practical, stable professions vs less marketable ones, but they should ultimately still make that decisions for themselves rather than be goaded into a particular direction.
This may very well backfire on Microsoft in two ways:
1) Cloud environments are much less dependent on Windows than desktops, and favour open source like Linux or BSD
2) If all applications become server side, then application compatibility on the desktop is no longer relevant, making the OS that desktop runs irrelevant
Failure in Cloud is not an option for Microsoft, and the days of their lock-in are numbered.
Actually, you can certainly reduce disk I/O with more RAM by caching frequently used data. Sure, it won't help if you need all different data all the time, but that's usually not the case.
Note that I qualified my statement to both include China's perception, and "proper".
In 1972, 1995, and in 2001, various Japanese prime ministers have issued what they considered to be a valid apology. Each time, China rejected the statement as a valid apology for one or more of three reasons: 1) the lack of the explicit mention of the word “apology,” 2) the lack of the explicit mention of China as the victim of Japanese aggression, and 3) the apology was only stated in a speech, but not written down in an official document. - See more at: http://www.tealeafnation.com/2012/12/has-japan-ever-apologized-to-china-for-its-wartime-aggression/#sthash.bIg5DWBO.dpuf
they believe their civilization has been unfairly held down for too long by hostile foreign powers and that it is finally time for their superior race/culture to take its rightful place of leadership on the world stage.
In China's case, there were the opium wars and the invasion by Japan. It doesn't help that China feels they have still never received a proper apology, and that there are Nanking deniers among Japan's right wing conservatives, including current prime minister Shinzo Abe.
China has also been in place of leadership throughout most of its history (so far as Asia is concerned at least), with the last few hundred years being the exception.
They had 75% of the smartphone market at a time when smartphones were a much smaller portion of the overall mobile phone market. Their drop has not been anywhere near as precipitous as the 96% drop you suggest.
Absolutely false. Ingress traffic is much cheaper than egress traffic. Try negotiating bandwidth pricing as an eyeball network (consumer ISP), vs as a hosting provider, and the difference in cost can be an order of magnitude. Peering is often based on equal ratios, and that applies for both egress and ingress traffic. When the traffic is balanced, the network that has more egress traffic may have to pay the other network for the difference. Therefore, ingress traffic not only costs these providers nothing, but can actually reduce their paid peering costs. As such, they will sell transit connections to eyeball networks for pennies on the Mb.
On the hosting side of things, data centers never purchase Microsoft Server licenses outright, as the actual instances of Windows Server are run by clients. Therefore, the only way to be compliant with Microsoft's terms is to offer licensing under the SPLA program, which offers licensing Microsoft Software for a monthly cost.
Also, with Windows Server 2012, they have gotten rid of Windows Server Enterprise, and there is now only really Windows Server Standard and Windows Server Data Center, which is functionally identical software. The only difference is that Standard allows for up to two instances on the same hardware, whereas Data Center Edition allows for unlimited instances on the same hardware. However, you need licenses for each physical processor. You can just add on additional Standard licenses until the Data Center Edition pricing makes sense. Perhaps one of the reason they're raising prices, is that with the increases in processing power per server, unlimited instances result in far many more virtuals now than before.
No, inbound port 80 for http was correct Although some ISP's do block outbound port 25 for smtp. I don't think any ISP would stay in business for long if they blocked outbound port 80 traffic.
What exact benefit does this supposedly 'open' laptop have over just buying something like a Thinkpad that uses Intel components that are well-supported by open source drivers on open source operating systems (Linux, *BSD, etc.)?
If it's to promote the use of standardized components that can be re-used in different laptop designs, it may serve to reduce costs or to increase the useful life of some of those parts. On the other hand, the standardization would also limit designs and prevent some cutting edge innovations from being utilized. All-in-all, as great as it's been to have interchangeable components on desktop builds, there's a reason why there's been limited standardization on laptops and servers, where innovation has more benefit. And considering how inexpensive laptops have already become simply due to competition, I see there being little to gain but much to lose from this approach.
Yes, Darwin does use a MACH kernel. But the MACH kernel itself is also derived from BSD, originally developed as additional code written directly into the 4.2BSD kernel.
Hosting their Internet facing servers at the office, behind a sonicwall firewall, is also a recipe for disaster in general let alone when being attacked. If they had hosted their servers in a proper data centre with DDoS mitigation services, the 'hackers' would have had a much harder time taking their servers offline.
Not nearly enough has been presented to demonstrate whether or not the technology is truly compelling or not. Is this purely a storage enclosure, or is there some type of cpu/mainboard for which this can be used as a high density, low cost SAN? Why type of bus/connectors/interfaces are used? Does this use fiber channel or 10Gb Ethernet? Can the full potential of all the drives' IO and IOPS be realized? What is the latency involved? Also, what's so specialized that the technology can be patented, and not easily replicated by a competitor?
Martha Stewart's prosecution was highly publicized so that it would appear that white collar criminals were being cracked down. If you're going to make the argument that this goes beyond an isolated case, and was a general and significant trend, please provide a citation.
Uhh, the Xeon E3 V2's which are the single socket Ivy Bridge server processors have been out for several months.
However, the Xeon E5 V2's i.e. Ivy Bridge-E(P|N) for 2 sockets aren't going to come out until Q3 2013, which is quite the delay from Ivy Bridge's initial release, and well after Haswell will be released.
Slashdot Mirror
I disagree. Progressive taxation simply aligns income tax more with disposable income than gross income. Ideally, there would be a straightforward way to calculate disposable income, but sadly there is not. As a percentage of disposable income, the poor are paying a disproportionately greater amount of tax than the wealthy. It also stands to reason that wealthier people are receiving greater benefit from societal structures, so it makes sense for them to make greater contributions to support those structures accordingly.
These are tenanted data centers, with customers on long-term contracts. It would take up to 3 years for contracts to expire before they could fully repurpose these facilities. Also, Verizon/Terremark owns the main interconnection point in Miami, including the Nap of The Americas peering exchange, where the vast majority of traffic to Latin America traverses through. If Microsoft were to simply repurpose that facility for internal use, they'd effectively cut off Latin America from the rest of the Internet. It's much more likely for a company in this market to be the acquirer. Someone like Equinix, Zayo, Internap, Coresite, or Cologix. 2.5B would be a lot for any one of these companies to shell out. There may also be some realty companies who specialize in leasing space for data centers like Digital Realty Trust who may be interested. Ultimately though, I think Verizon may need to break up some of these assets if no one company can pony up the entire sum.
One big, big name is Youtube, who was burning through other people's money faster than a drunk Kennedy and getting rightfully sued every 5 minutes for copyright infringement. They had a cool idea, and a completely non-sustainable business model that was guaranteed to put them belly-up within 36 months until Google bought them. Google brought to bear their expertise in funding a free service in a way that keeps customers happy (aka the best targeted advertising available) , allowing YouTube to survive and thrive rather than burning away investors' money until investors got sick of it and'the whole thing imploded.
YouTube's business model from the beginning was to get bought by Google, and so they focused their business towards that end and succeeded. Had that not been their primary goal, they may have done things very differently. In any event, using them as an example of a smaller company that needed to be bought isn't wrong, but the assertion that they needed the good processes that came from the larger company is false. They were very efficient at exiting for maximum return for their shareholders.
The issue is not about ISP's being forced to peer with content providers like Netflix whether they like it or not. It's the major eyeball networks like Comcast and AT&T deliberately avoiding upgrading peering capacity, so that they can strong-arm Netflix into paid peering agreements instead. Let's not forget that Comcast is also trying to pressure Tier 1 transit providers like Level3 to pay for access to their eyeballs now, even though the miles of fiber in the ground that they operate, and thus their share of transport costs, pale in comparison. As a smaller provider, you should most certainly want to peer with any content network with a significant amount of traffic to your network, as you would have to pay for transit otherwise. There is no way you'd be paying more for cross-connects than you would for transit, and the equipment costs are irrelevant because you're doing that same traffic regardless, whether it comes through a transit or a peer. Your traffic levels aren't going to magically increase when you start peering, unless your existing transit is congested, which is something you should most definitely want to avoid.
Netflix does most of its heavy lifting through its own CDN boxes running FreeBSD + nginx, which it will place on ISPs' networks for free, although the ISP would be responsible for the space and power. They most certainly do have their own data centres though, with infrastructure that goes well beyond Amazon.
I call BS. We're at the Westin as well, and x-connects in the 19th floor meet-me room are free, while x-connects between floors are a one-time build cost. even if you don't have your own space and are leasing lines from a data centre, they're only $50-$200/mo. Powered equipment may not be permitted in the meet-me room, but there is nothing that would stop you from putting some passive DWDM muxes in there so you could even run 40-80x 10Gb-100Gb links over the same fiber. The only way I see x-connects being expensive is you're using Equinix, but that would be your own business decision, and not inherent costs to peering.
Also, 10Gb optics sourced directly from China where they're manufactured can be purchased for less than $100 for LR optics, or less than $1000 for DWDM optics, which is peanuts for a one-time cost. Sure, line cards and therefore physical interfaces are expensive, but that's why public peering exchanges like the SIX (Seattle Internet Exchange) exist so you can peer with many different networks over the same interface. If you are doing enough traffic over a link to justify private peering however, your equipment costs should be built-in to your business model, and whether it's peering or other traffic shouldn't matter.
Although I do agree with your assessment that this isn't a net neutrality issue, it most certainly is a case of large eyeball networks trying to double-dip on charging for bandwidth, and criticism of them is well deserved.
This. Instead of pushing young people towards a certain path, and converting highschools and earlier into trade schools, why not just give them the best, all-rounded education possible and allow them to decide for themselves what they want to do? That's not to say we shouldn't teach them the value of more practical, stable professions vs less marketable ones, but they should ultimately still make that decisions for themselves rather than be goaded into a particular direction.
This may very well backfire on Microsoft in two ways:
1) Cloud environments are much less dependent on Windows than desktops, and favour open source like Linux or BSD
2) If all applications become server side, then application compatibility on the desktop is no longer relevant, making the OS that desktop runs irrelevant
Failure in Cloud is not an option for Microsoft, and the days of their lock-in are numbered.
Actually, you can certainly reduce disk I/O with more RAM by caching frequently used data. Sure, it won't help if you need all different data all the time, but that's usually not the case.
Note that I qualified my statement to both include China's perception, and "proper".
In 1972, 1995, and in 2001, various Japanese prime ministers have issued what they considered to be a valid apology. Each time, China rejected the statement as a valid apology for one or more of three reasons: 1) the lack of the explicit mention of the word “apology,” 2) the lack of the explicit mention of China as the victim of Japanese aggression, and 3) the apology was only stated in a speech, but not written down in an official document. - See more at: http://www.tealeafnation.com/2012/12/has-japan-ever-apologized-to-china-for-its-wartime-aggression/#sthash.bIg5DWBO.dpuf
they believe their civilization has been unfairly held down for too long by hostile foreign powers and that it is finally time for their superior race/culture to take its rightful place of leadership on the world stage.
In China's case, there were the opium wars and the invasion by Japan. It doesn't help that China feels they have still never received a proper apology, and that there are Nanking deniers among Japan's right wing conservatives, including current prime minister Shinzo Abe. China has also been in place of leadership throughout most of its history (so far as Asia is concerned at least), with the last few hundred years being the exception.
False. The US would be near the top of the list. http://www.wto.org/english/tratop_e/dispu_e/dispu_status_e.htm Most other countries don't have the political might to get away with what the US is able to.
They had 75% of the smartphone market at a time when smartphones were a much smaller portion of the overall mobile phone market. Their drop has not been anywhere near as precipitous as the 96% drop you suggest.
Absolutely false. Ingress traffic is much cheaper than egress traffic. Try negotiating bandwidth pricing as an eyeball network (consumer ISP), vs as a hosting provider, and the difference in cost can be an order of magnitude. Peering is often based on equal ratios, and that applies for both egress and ingress traffic. When the traffic is balanced, the network that has more egress traffic may have to pay the other network for the difference. Therefore, ingress traffic not only costs these providers nothing, but can actually reduce their paid peering costs. As such, they will sell transit connections to eyeball networks for pennies on the Mb.
On the hosting side of things, data centers never purchase Microsoft Server licenses outright, as the actual instances of Windows Server are run by clients. Therefore, the only way to be compliant with Microsoft's terms is to offer licensing under the SPLA program, which offers licensing Microsoft Software for a monthly cost. Also, with Windows Server 2012, they have gotten rid of Windows Server Enterprise, and there is now only really Windows Server Standard and Windows Server Data Center, which is functionally identical software. The only difference is that Standard allows for up to two instances on the same hardware, whereas Data Center Edition allows for unlimited instances on the same hardware. However, you need licenses for each physical processor. You can just add on additional Standard licenses until the Data Center Edition pricing makes sense. Perhaps one of the reason they're raising prices, is that with the increases in processing power per server, unlimited instances result in far many more virtuals now than before.
FreeBSD Graphics Driver Download Version: 1.0-3203 File Size: 3 MB Release Date: November 7, 2002
And pretty close to a decade already with OpenOffice: http://www.freshports.org/editors/openoffice-1.1/
11 Mar 2004 12:36:03
Just because many people use it incorrectly does not make that usage correct.
No, inbound port 80 for http was correct Although some ISP's do block outbound port 25 for smtp. I don't think any ISP would stay in business for long if they blocked outbound port 80 traffic.
What exact benefit does this supposedly 'open' laptop have over just buying something like a Thinkpad that uses Intel components that are well-supported by open source drivers on open source operating systems (Linux, *BSD, etc.)? If it's to promote the use of standardized components that can be re-used in different laptop designs, it may serve to reduce costs or to increase the useful life of some of those parts. On the other hand, the standardization would also limit designs and prevent some cutting edge innovations from being utilized. All-in-all, as great as it's been to have interchangeable components on desktop builds, there's a reason why there's been limited standardization on laptops and servers, where innovation has more benefit. And considering how inexpensive laptops have already become simply due to competition, I see there being little to gain but much to lose from this approach.
Yes, Darwin does use a MACH kernel. But the MACH kernel itself is also derived from BSD, originally developed as additional code written directly into the 4.2BSD kernel.
Hosting their Internet facing servers at the office, behind a sonicwall firewall, is also a recipe for disaster in general let alone when being attacked. If they had hosted their servers in a proper data centre with DDoS mitigation services, the 'hackers' would have had a much harder time taking their servers offline.
Not nearly enough has been presented to demonstrate whether or not the technology is truly compelling or not. Is this purely a storage enclosure, or is there some type of cpu/mainboard for which this can be used as a high density, low cost SAN? Why type of bus/connectors/interfaces are used? Does this use fiber channel or 10Gb Ethernet? Can the full potential of all the drives' IO and IOPS be realized? What is the latency involved? Also, what's so specialized that the technology can be patented, and not easily replicated by a competitor?
Martha Stewart's prosecution was highly publicized so that it would appear that white collar criminals were being cracked down. If you're going to make the argument that this goes beyond an isolated case, and was a general and significant trend, please provide a citation.
Uhh, the Xeon E3 V2's which are the single socket Ivy Bridge server processors have been out for several months. However, the Xeon E5 V2's i.e. Ivy Bridge-E(P|N) for 2 sockets aren't going to come out until Q3 2013, which is quite the delay from Ivy Bridge's initial release, and well after Haswell will be released.
It's Jordan Hubbard, not John Hubbard. He was one of the original founders of FreeBSD. http://en.wikipedia.org/wiki/Jordan_Hubbard http://www.turbofuzz.com/jkh/