It sounds, though, that your performance gains come from moving swap to a separate drive, not from using a file on a FAT32 partition rather than a swap partition.
Or you could just do the normal thing and use a swap partition. Are there any actual benefits to using a swap file on a file system instead of an actual swap partition?
I mean, sure, it's incredibly useful in certain circumstances when you CAN'T make a swap partition. My VPS has one big partition, stored on a SAN, and I can't repartition anything. Being able to make a swap file was super handy. But in a real computer, this shouldn't be a concern.
Except that there are only two main carriers in most parts of Canada. There is Rogers, and Bell. All others are MVNOs or subsidiaries of those two.
Take Montreal. We have tons of cell providers. All function on one of those two networks. Telus is an MVNO for Bell (and vice versa out west). Fido is owned by Rogers. Virgin Mobile is an MVNO of Bell. Koodo is owned by Telus, who is in turn an MVNO of Bell in Ontario/Quebec. Solo is owned by Bell. And so on.
So, really, we have one GSM network, and one CDMA network. Bell may have more MVNOs, but that doesn't say much about the network.
True, but this is still a big improvement over existing solutions. Given higher resolution photos (as technology marches onward), eventually these types of things will get pretty darned accurate.
An infinite number of monkeys given an infinite amount of time would more likely produce infinitely many pages filled with the letter "e". Apparently they like to hold down the buttons.
Hydro Quebec generates 35.19 GW, and has a population of 7,546,131 living in 3,452,300 private dwellings.
That comes out to about 10.2 KW per home, including supporting industries.
Therefore, by the Quebec figures, 4 GW would produce enough power for 392 thousand homes and supporting industries.
Another factor to consider is that he is investing billions plural, not singular. If we assume billions means 2+, that means the cost per home is, by actual observed figures, over $5100.
Of course, I'm ignoring supporting industries; businesses and companies pay for their own power. But they still need power, and must be counted in figures. If you say there are X homes in an area, you can't say you need to power X homes to have something as your exclusive power source; where are all the businesses going to get power?
While Quebec's hydro-only policy (with small exceptions, about 3% comes from non-clean sources) worked well for us, that doesn't work in areas that don't have the same hydro capacity. If you consider that Quebec's power is generated about 1000KM away from the biggest population centre (Montreal), distance isn't a terribly large factor. For this reason, I'm convinced that a combination of technologies is the best solution.
Wind is a fairly constant energy source, as it's always blowing somewhere. Solar only generates power during the day, but that also happens to be peak time. Hydro can produce an enormous amount of power consistently year-round, but is very location-specific. Nuclear is safe and clean, but expensive.
Combine all these technologies along with good long-distance transmission techniques and you produce a very effective and clean power system for the entire US.
Heck, Quebec is covering 1.37 million square kilometers of land with a low-emission power source...
There are a few things that come to mind. Not all of them may necessarily have any benefit.
1) Bond two or more dialup modems. Find an ISP that supports MLPPP for dialup modems. Typically requires a second phone line, which is not usually super expensive.
2) Stick to only hardware modems (not WinModems). While this may not be true today, last time I was using dialup, winmodems tended to add 50-100ms of extra latency.
3) Use a local caching proxy. This provides a greater benefit when there are more than one computers being used, although I doubt that's the case here. Still, this is better than a browser cache, which might expire older data that is still perfectly valid (ex: slashdot's logo, if you visit once a month; it doesn't change very often).
4) Use a remote "accelerator" proxy. I can't point you to any specific examples. Such things typically compress text (not sure if that's more efficient than dialup's hardware compression), recompress images (gif->png, jpeg with lower quality), and reduce the impact of multiple round-trips (if you choose one that is close to your ISP).
If you combine all these steps, you'll have an actual 112kbit connection that feels more like a 256kbit connection when surfing the web. It's still not quite fast enough for stuff like YouTube, though. You'll need at least four bonded lines to do that.
They did make the heatsink easily removable, but the drive is designed for the 15mm enterprise form factor (servers, for example), not laptop form factors.
The heatsink (which reduces average temperatures by 5-7 degrees) does work (it's not for show), but these things will never go in laptops.
Not to mention the fact that 3 gigawatts isn't all that much. Here in Quebec, our hydro company has an installed capacity of over 30 gigawatts, and we've only got a population of just under 8 million. Of course, we sell a good deal of it to Ontario and the US, since we've got a big surplus...
It's great that they can do all this physics stuff on the GPU, but they seem to be forgetting the other thing you need that for; graphics. Games like Crysis don't exactly leave room on the GPU to do anything but graphics, and throwing physics into the mix will only make things worse.
My DSL ISP's base cost of $30 CAD includes 200GB (at 5mbit) transfer, increasing cap is $10/100GB, or $0.25 per GB overage. Bandwidth usage is averaged over two months to help smooth out spikes.
Bandwidth isn't that expensive. You're getting overcharged.
You'd detect the failure the instant you tried to read the sector, and could speed up this process by using idle time to check sectors for corruption (checksums help). With double parity, you'd have to have the same stripe go bad on three different disks to lose data, and I have trouble believing that the failures would spread that fast.
I mean, I still wouldn't trust it in a production environment, as you said, I just wonder how a more dynamic system (RAID-Z, for example) would handle the same scenario. At the very least if you decided to simply try to get one clean copy of all data off the disks to migrate it to backups to prevent data loss.
As in, your test would be constantly reading all data on the disk in a loop, and seeing how long it would take before data was lost permanently.
Unfortunately, it's extremely expensive and all reviews point to abysmal performance. Also, doesn't even have an ethernet jack to act as NAS. Sort of kills the deal, don't you think?
If, for that price (or less, those things are really overpriced), that had onboard GigE and fixed their performance problems, it'd probably be a decent deal.
The thing is, the big initial cost is a one-time thing. After that, you'll have set down a network with virtually unlimited scalability (limited only by what equipment you put on either end of the strands), and a lifetime that should be at least a few decades.
That, and didn't Americans already pay for all this through tax money years ago? I seem to recall some big stink about the telecoms having been given hundreds of billions of dollars to build out these networks.
Independent ISPs can circumvent the issue by colocating their own DSLAMs in Bell's COs. At that point, Bell can't interfere, because the ISPs get (and are paying for) access to the raw, dumb copper. Of course, Bell doesn't care about throttling at that point, because Bell's network isn't used for backhauling (although it can be, but you're buying LAN extensions then).
Some ISPs do this already. The cost of serving a large number of users, though, is prohibitive. Some ISPs like Primus use a hybrid approach, a mix of their own DSLAMs and Bell's. This gives them the best of both worlds.
If an ISP were to try to exclusively colocate their own DSLAMs, they wouldn't get access to Bell's remotes. My understanding is that Bell was lucky to get those remotes at all; it's now virtually impossible to get right-of-way to install a remote, and the independent ISPs (even if they formed a coalition and shared DSLAMs and remotes) would never be able to get anywhere installing remotes.
There IS no alternative to Bell, and no possibility for independent ISPs to become completely isolated from Bell's network. The best they can do is move some of their customers off Bell's equipment. The only alternative is cable, but the cable market is far more fragmented; there are at least three cablecos in Bell's Ontario/Quebec territory alone (Rogers, Cogeco, Videotron). It would be far more difficult to manage. On top of that, my understanding is that the restrictions aren't as tight on wholesold cable; Videotron applies their own caps to wholesalers as well despite the fact the wholesalers are providing their own upstream transit providers!
They currently service Alberta, British Columbia, Ontario, and Quebec. They're shortly going to expand to Bell Aliant territory, which would include (I believe) Nova Scotia, Prince Edward Island, New Brunswick and Newfoundland and Labrador. At that point, TekSavvy will service eight provinces.
Yes, because it's a game of averages. TekSavvy pays about 3 cents per megabit for the unmetered customers. They're betting that the average customer uses less than 500GB/mth (their estimated break-even point at the $40 pricepoint). And since the average unlimited user used 118.47GB in January (please see the DSLR thread where TekSavvy's owner listed their average user bandwidth figures: http://www.dslreports.com/forum/r20029507-Rocky-DecJan-bandwidth-stats), that game of averages currently works in TekSavvy's favour.
They used to charge $30/mth for unlimited service. Their breakeven point then would have been about 167GB. They were getting dangerously close to that breakeven point (and will probably pass it in the future), so were forced to raise the price to $40/mth.
This is based on the fact that Bell charges $20.50/mth per customer, and we usually guess that TekSavvy's other expenses are covered by the other $4.50 or so, and so we (the customers) usually say that $25/mth is spoken for, and they have the other $5 (or $15 in the case of the unlimited service) to pay for bandwidth and make a profit.
It depends how you define overselling. All ISPs oversell. The entire internet is oversold. It's the concept that makes it financially viable.
If I have 100 customers who each have 5mbit connections, with average usage of 50mbit/s and peak usage of 75mbit/s, why should I (as an ISP) pay for 500mbit/s?
Overselling only becomes bad when you don't have enough bandwidth to handle peak loads. The internet and ISPs cannot function without overselling, and that includes TekSavvy.
No, TekSavvy does have a cap, and does charge for overage. Of course, your $30/mth DSL line has a 200GB/mth cap, and overage is $0.10 to $0.25 (depending on if you pre-purchase a 100GB chunk or not). So they're really not discouraging heavy users, just trying to get a handle on the insanely-high users. Those users are encouraged to sign up for the unmetered cogent-only service for $40/mth. Same speeds, still have the ability to saturate your line, except you're on cogent instead of the normal multi-homed "premium" network.
The reseller doesn't OWN A to B, but they're paying for it.
Some wholesalers actually do own A to B. The problem is that in order to do that, you need to colocate your own DSLAMs in the COs, which limits your coverage and reach. It does have the advantage of letting you pick your own profiles/technology (ADSL, ADSL2+, VDSL, etc) since the wholesaler has access to the raw copper wire, and have zero interference since you're providing your own backhaul connection from the CO.
Slashdot Mirror
It sounds, though, that your performance gains come from moving swap to a separate drive, not from using a file on a FAT32 partition rather than a swap partition.
Or you could just do the normal thing and use a swap partition. Are there any actual benefits to using a swap file on a file system instead of an actual swap partition?
I mean, sure, it's incredibly useful in certain circumstances when you CAN'T make a swap partition. My VPS has one big partition, stored on a SAN, and I can't repartition anything. Being able to make a swap file was super handy. But in a real computer, this shouldn't be a concern.
So, a multi-million dollar industry is "basically zero"? You have a funny definition of zero.
Except that there are only two main carriers in most parts of Canada. There is Rogers, and Bell. All others are MVNOs or subsidiaries of those two.
Take Montreal. We have tons of cell providers. All function on one of those two networks. Telus is an MVNO for Bell (and vice versa out west). Fido is owned by Rogers. Virgin Mobile is an MVNO of Bell. Koodo is owned by Telus, who is in turn an MVNO of Bell in Ontario/Quebec. Solo is owned by Bell. And so on.
So, really, we have one GSM network, and one CDMA network. Bell may have more MVNOs, but that doesn't say much about the network.
The iPhone will only put the "lackluster Java-based cell phone gaming market to death" when most phone users out there are iPhone users.
Apple has captured an impressive portion of the smartphone market, but their overall market share among all cellphones is minuscule.
True, but this is still a big improvement over existing solutions. Given higher resolution photos (as technology marches onward), eventually these types of things will get pretty darned accurate.
How is being illegally refused in 75% of requests considered "[not] too bad"?
An infinite number of monkeys given an infinite amount of time would more likely produce infinitely many pages filled with the letter "e". Apparently they like to hold down the buttons.
They're not right.
Hydro Quebec generates 35.19 GW, and has a population of 7,546,131 living in 3,452,300 private dwellings.
That comes out to about 10.2 KW per home, including supporting industries.
Therefore, by the Quebec figures, 4 GW would produce enough power for 392 thousand homes and supporting industries.
Another factor to consider is that he is investing billions plural, not singular. If we assume billions means 2+, that means the cost per home is, by actual observed figures, over $5100.
Of course, I'm ignoring supporting industries; businesses and companies pay for their own power. But they still need power, and must be counted in figures. If you say there are X homes in an area, you can't say you need to power X homes to have something as your exclusive power source; where are all the businesses going to get power?
While Quebec's hydro-only policy (with small exceptions, about 3% comes from non-clean sources) worked well for us, that doesn't work in areas that don't have the same hydro capacity. If you consider that Quebec's power is generated about 1000KM away from the biggest population centre (Montreal), distance isn't a terribly large factor. For this reason, I'm convinced that a combination of technologies is the best solution.
Wind is a fairly constant energy source, as it's always blowing somewhere. Solar only generates power during the day, but that also happens to be peak time. Hydro can produce an enormous amount of power consistently year-round, but is very location-specific. Nuclear is safe and clean, but expensive.
Combine all these technologies along with good long-distance transmission techniques and you produce a very effective and clean power system for the entire US.
Heck, Quebec is covering 1.37 million square kilometers of land with a low-emission power source...
There are a few things that come to mind. Not all of them may necessarily have any benefit.
1) Bond two or more dialup modems. Find an ISP that supports MLPPP for dialup modems. Typically requires a second phone line, which is not usually super expensive.
2) Stick to only hardware modems (not WinModems). While this may not be true today, last time I was using dialup, winmodems tended to add 50-100ms of extra latency.
3) Use a local caching proxy. This provides a greater benefit when there are more than one computers being used, although I doubt that's the case here. Still, this is better than a browser cache, which might expire older data that is still perfectly valid (ex: slashdot's logo, if you visit once a month; it doesn't change very often).
4) Use a remote "accelerator" proxy. I can't point you to any specific examples. Such things typically compress text (not sure if that's more efficient than dialup's hardware compression), recompress images (gif->png, jpeg with lower quality), and reduce the impact of multiple round-trips (if you choose one that is close to your ISP).
If you combine all these steps, you'll have an actual 112kbit connection that feels more like a 256kbit connection when surfing the web. It's still not quite fast enough for stuff like YouTube, though. You'll need at least four bonded lines to do that.
They did make the heatsink easily removable, but the drive is designed for the 15mm enterprise form factor (servers, for example), not laptop form factors.
The heatsink (which reduces average temperatures by 5-7 degrees) does work (it's not for show), but these things will never go in laptops.
I don't know what they were smoking when they wrote that proposal, but there IS a copper spec. IEEE 802.3an-2006 (yes, 2006), or 10GBase-T.
Rated for 55m over Cat6 cable, or the full 100m over Cat6a cable.
Your Cat6 wiring is non-optimal for 10GigE, but will at least work.
Not to mention the fact that 3 gigawatts isn't all that much. Here in Quebec, our hydro company has an installed capacity of over 30 gigawatts, and we've only got a population of just under 8 million. Of course, we sell a good deal of it to Ontario and the US, since we've got a big surplus...
It's great that they can do all this physics stuff on the GPU, but they seem to be forgetting the other thing you need that for; graphics. Games like Crysis don't exactly leave room on the GPU to do anything but graphics, and throwing physics into the mix will only make things worse.
My DSL ISP's base cost of $30 CAD includes 200GB (at 5mbit) transfer, increasing cap is $10/100GB, or $0.25 per GB overage. Bandwidth usage is averaged over two months to help smooth out spikes.
Bandwidth isn't that expensive. You're getting overcharged.
You'd detect the failure the instant you tried to read the sector, and could speed up this process by using idle time to check sectors for corruption (checksums help). With double parity, you'd have to have the same stripe go bad on three different disks to lose data, and I have trouble believing that the failures would spread that fast.
I mean, I still wouldn't trust it in a production environment, as you said, I just wonder how a more dynamic system (RAID-Z, for example) would handle the same scenario. At the very least if you decided to simply try to get one clean copy of all data off the disks to migrate it to backups to prevent data loss.
As in, your test would be constantly reading all data on the disk in a loop, and seeing how long it would take before data was lost permanently.
Unfortunately, it's extremely expensive and all reviews point to abysmal performance. Also, doesn't even have an ethernet jack to act as NAS. Sort of kills the deal, don't you think?
If, for that price (or less, those things are really overpriced), that had onboard GigE and fixed their performance problems, it'd probably be a decent deal.
The thing is, the big initial cost is a one-time thing. After that, you'll have set down a network with virtually unlimited scalability (limited only by what equipment you put on either end of the strands), and a lifetime that should be at least a few decades.
That, and didn't Americans already pay for all this through tax money years ago? I seem to recall some big stink about the telecoms having been given hundreds of billions of dollars to build out these networks.
Independent ISPs can circumvent the issue by colocating their own DSLAMs in Bell's COs. At that point, Bell can't interfere, because the ISPs get (and are paying for) access to the raw, dumb copper. Of course, Bell doesn't care about throttling at that point, because Bell's network isn't used for backhauling (although it can be, but you're buying LAN extensions then).
Some ISPs do this already. The cost of serving a large number of users, though, is prohibitive. Some ISPs like Primus use a hybrid approach, a mix of their own DSLAMs and Bell's. This gives them the best of both worlds.
If an ISP were to try to exclusively colocate their own DSLAMs, they wouldn't get access to Bell's remotes. My understanding is that Bell was lucky to get those remotes at all; it's now virtually impossible to get right-of-way to install a remote, and the independent ISPs (even if they formed a coalition and shared DSLAMs and remotes) would never be able to get anywhere installing remotes.
There IS no alternative to Bell, and no possibility for independent ISPs to become completely isolated from Bell's network. The best they can do is move some of their customers off Bell's equipment. The only alternative is cable, but the cable market is far more fragmented; there are at least three cablecos in Bell's Ontario/Quebec territory alone (Rogers, Cogeco, Videotron). It would be far more difficult to manage. On top of that, my understanding is that the restrictions aren't as tight on wholesold cable; Videotron applies their own caps to wholesalers as well despite the fact the wholesalers are providing their own upstream transit providers!
Bell killed off the $25/mth unlimited plan, and will likely remove it from customers like you.
They're also getting rid of the $30/mth overage maximum as of June, although that's only on new customers. For now.
I doubt you'll be in as good a mood when you're throttled and paying $1.50 per gig over your 30/60 gig cap.
They currently service Alberta, British Columbia, Ontario, and Quebec. They're shortly going to expand to Bell Aliant territory, which would include (I believe) Nova Scotia, Prince Edward Island, New Brunswick and Newfoundland and Labrador. At that point, TekSavvy will service eight provinces.
Yes, because it's a game of averages. TekSavvy pays about 3 cents per megabit for the unmetered customers. They're betting that the average customer uses less than 500GB/mth (their estimated break-even point at the $40 pricepoint). And since the average unlimited user used 118.47GB in January (please see the DSLR thread where TekSavvy's owner listed their average user bandwidth figures: http://www.dslreports.com/forum/r20029507-Rocky-DecJan-bandwidth-stats), that game of averages currently works in TekSavvy's favour.
They used to charge $30/mth for unlimited service. Their breakeven point then would have been about 167GB. They were getting dangerously close to that breakeven point (and will probably pass it in the future), so were forced to raise the price to $40/mth.
This is based on the fact that Bell charges $20.50/mth per customer, and we usually guess that TekSavvy's other expenses are covered by the other $4.50 or so, and so we (the customers) usually say that $25/mth is spoken for, and they have the other $5 (or $15 in the case of the unlimited service) to pay for bandwidth and make a profit.
It depends how you define overselling. All ISPs oversell. The entire internet is oversold. It's the concept that makes it financially viable.
If I have 100 customers who each have 5mbit connections, with average usage of 50mbit/s and peak usage of 75mbit/s, why should I (as an ISP) pay for 500mbit/s?
Overselling only becomes bad when you don't have enough bandwidth to handle peak loads. The internet and ISPs cannot function without overselling, and that includes TekSavvy.
No, TekSavvy does have a cap, and does charge for overage. Of course, your $30/mth DSL line has a 200GB/mth cap, and overage is $0.10 to $0.25 (depending on if you pre-purchase a 100GB chunk or not). So they're really not discouraging heavy users, just trying to get a handle on the insanely-high users. Those users are encouraged to sign up for the unmetered cogent-only service for $40/mth. Same speeds, still have the ability to saturate your line, except you're on cogent instead of the normal multi-homed "premium" network.
The reseller doesn't OWN A to B, but they're paying for it.
Some wholesalers actually do own A to B. The problem is that in order to do that, you need to colocate your own DSLAMs in the COs, which limits your coverage and reach. It does have the advantage of letting you pick your own profiles/technology (ADSL, ADSL2+, VDSL, etc) since the wholesaler has access to the raw copper wire, and have zero interference since you're providing your own backhaul connection from the CO.