Instead of focusing primarily on fixing vulnerabilities, businesses should turn toward deterring threats, including detecting attacks and responding to them, he said. There have to be penalties for attackers, Chabinsky added.
The problem with the logic here is, a company is trying to protect data that is worth, at least to them, possibly millions of dollars. The attacker can be using a crappy dell system and maybe a bot-net he acquired (somehow,) for a total cost of fuck all. Even if the company can respond and make the attackers gear explode (and really we are being very pie in the sky there aren't we) your still only inconveniencing the attacker fuck all. Even best case for the company with this response situation there is still very little for the attacker to loose for a possible great gain or great net result if the aim is sabotage. Moral problems aside, this does not make economic sense in the end, there is no deterrent in most cases.
The only recourse for them is to prevent the attack (i.e. fixing vulnerabilities) and report breaches to the authorities. Unless responses include international bounty hunters?
I agree, it would be much better to just make the content available to everyone and be done with it. There is a but to this as things are organized at the moment. Currently, it would cost the BBC a bomb to distribute the content, via their iPlayer web things, to everyone, even if we are talking the EU and not the world. It would divert funds form producing content to coping with distribution.
What would make more sense to me is if someone like the ISP would do the distribution - i.e. BBC would give the content to the content to each ISP who is interested in distributing to their customers - all the bandwidth issues then become the issue of the ISP. This makes more good logical sense that an ISP would be in the better position to cater to bandwidth issues and the content producer can concentrate on production. This is far from how the world works at the moment though, major changes in the way we think about this is required.
The other model that makes sense is for all this content to be torrent based - this automatically distributes the bandwidth load where it is being used (assuming people don't leach too much! maintain a ratio etc.) This requires an even more radical rethink of how the content is thought of currently...
In short, there is, with it's current centralized distribution model, reason for bodies like the BBC to geographically restrict content. You need to de-centralise the distribution model for it to be fair otherwise. People don't think of it all like that right now though.
Hmm, that's kind of piddle and crap compare to what I was looking at today, one department have over a million folders (24 terabytes) don't know how many files yet: taking too long. I'd be suspicious of performance scaling well to that sort of level.
Highly available? How do you do that? I don't want remote backups, I want a highly available system that can just start using a second node transparently when one goes down. DFS/VSS does that for you.
ZFS looks great for snapshots... does it do anything in a distributed manner? That is, can it's storage pools be on different physical systems? how can you do ZFS without a single point of failure? is it possible?
Do any of these options work without a single point of failure? Ceph by itself sounds awesome, but how do you get versioning of some sort on top of it without introducing a single point of failure? I guess you could stick load balances in front of things like the WebDAV option, but that is getting very complicated. DFS/VSS does quite easily introduce a system without a single point of failure and is still looking like it's mighty useful compared to all available in the Linux world.
I think you present the first real options of all the posts here but they are not easy options.
What is the performance of this sort of thing like? I'm thinking of the case of a business using this constantly. Is it going to work for a few hundred users using a share daily? Or is that just going to make it die? I was thinking something that was a copy-on-write type thing, not a slow cron job type thing, but if it does perform OK...
Well, to be more specific the conductor facilitates a single interpretation in the change of pulse and other variables open to interpretation. A good orchestra can easily keep a steady pulse and play together rhythmically without a conductor just fine. They can even start together blindfolded, this is about listening to each other, esp breathing, it is actually not as hard as you might think. A very good orchestra can even come to a good consensus as to musical interpretation without a conductor, but will generally come together much faster with a good conductor. A very good orchestra will completely ignore a bad conductor during a performance and sound better for it.
School orchestras and the like will have conductors perform a more "keep everyone together rhythmically" type function, but this is not the ideal situation. More of an aid to learning situation.
An independent! Parties suck because they are parties.
The greens are not so bad. They used to be radical hippies but they grew up, they would be my next choice after an independent. They do suffer from that past image still, but their policies now look the most sensible of any party.
Unless they are trying to find a free channel for data and there just isn't one because 6 people around you are doing this all on different channels. Allocation of another part of the spectrum for power transmission: much better idea than trying to use the same ones as for data.
Using incidental energy that is being used for data transmission anyway - nothing wrong with that. Flooding data channels with noise, not so nice for other people around you.
"No Party may require a service supplier, as a condition for supplying a service or investing in its territory, to: (a) use computing facilities located in the Party’s territory."
So my reasoning for not using your USA located computing facilities is not because the are in the USA, it is because you can not grantee the level of data security I require at that facility. The fact that this happens to be because the facility is located in a particular territory with stupid laws - relevant but NOT the end reason I'm refusing to deal with it. The reason is security requirements I have, not physical location. Not a problem?
More info on the specs of the "supercomputer" that TFA only glossed over:
The result is the custom-built supercomputer, which we call Minwa . It is comprised of 36 server nodes, each with 2 six-core Intel Xeon E5-2620 processors. Each sever con- tains 4 Nvidia Tesla K40m GPUs and one FDR InfiniBand (56Gb/s) which is a high-performance low-latency inter- connection and supports RDMA. The peak single precision floating point performance of each GPU is 4.29TFlops and each GPU has 12GB of memory. Thanks to the GPUDirect RDMA, the InfiniBand network interface can access the re- mote GPU memory without involvement from the CPU. All the server nodes are connected to the InfiniBand switch. Figure 1 shows the system architecture. The system runs Linux with CUDA 6.0 and MPI MVAPICH2, which also enables GPUDirect RDMA.
In total, Minwa has 6.9TB host memory, 1.7TB device memory, and about 0.6PFlops theoretical single precision peak performance.
It's not that powerful overall, but seems to be well thought out for what it is doing. I do the see point about fudging data somehow, they do provide a lot of information of what they supposedly did here
I don't know how this is verifiable, it's not like they have released source code or binaries for the software as far as I can tell.
Australia fixes most of the problems you describe simply: voting is mandatory, you must give time for employees to go and vote or face very stiff fines.
ARIA produces some "top charts" in Australia, here's how.
This is much better than what they used to do many years ago: rely on distribution figures (that is, how many unit were shipped this week.) I still don't think they reflect much except the skills of advertisers.
That is all true, but the economic implications for some situations make trading off privacy and control worth while. If you are not a tech savvy individual or company large enough to make the economy of scale worth while for your own IT department SAS can be a good thing economically.
Think about the situation of a small business with 10 employees. You want the ten employees to have company email and be able to share documents with each other. Do you:
1) buy a server, a backup solution, buy all the software required, engage a consulting company to get it all going for you and say support to said company when stuff breaks 2) buy a server, a backup solution, buy all the software required, hire an IT guy part time forever (assuming you can find a good one who will work part time) 3)pay $5 a month per user (Australian price, not sure about elsewhere) for google apps and gmail and be done with it.
Yes, option three trades away control to a large extent and privacy to some extent, but does the small business care? Unlikely they do. Should they care? Another question all together, maybe they should some times. The convenience and economic benefits (i.e. cheap) are too good to pass over in this small business situation. Should a large company of 20,000 employees do the same? I don't think so, having your own infrastructure at that size makes a lot more sense.
As to the original question, I can see a place for both thick and think clients in the future. You have to think about who your end client is going to be and a product that suits one sort of client. One size does not fit all and one product is not going to dominate all of a market because it is thick or thin client. I don't think there will ever be a situation in the future where all, say, word-processors are either thick or thin clients because this is always what is best.
Slashdot Mirror
From TFA:
Instead of focusing primarily on fixing vulnerabilities, businesses should turn toward deterring threats, including detecting attacks and responding to them, he said. There have to be penalties for attackers, Chabinsky added.
The problem with the logic here is, a company is trying to protect data that is worth, at least to them, possibly millions of dollars. The attacker can be using a crappy dell system and maybe a bot-net he acquired (somehow,) for a total cost of fuck all. Even if the company can respond and make the attackers gear explode (and really we are being very pie in the sky there aren't we) your still only inconveniencing the attacker fuck all. Even best case for the company with this response situation there is still very little for the attacker to loose for a possible great gain or great net result if the aim is sabotage. Moral problems aside, this does not make economic sense in the end, there is no deterrent in most cases.
The only recourse for them is to prevent the attack (i.e. fixing vulnerabilities) and report breaches to the authorities. Unless responses include international bounty hunters?
I agree, it would be much better to just make the content available to everyone and be done with it. There is a but to this as things are organized at the moment. Currently, it would cost the BBC a bomb to distribute the content, via their iPlayer web things, to everyone, even if we are talking the EU and not the world. It would divert funds form producing content to coping with distribution.
What would make more sense to me is if someone like the ISP would do the distribution - i.e. BBC would give the content to the content to each ISP who is interested in distributing to their customers - all the bandwidth issues then become the issue of the ISP. This makes more good logical sense that an ISP would be in the better position to cater to bandwidth issues and the content producer can concentrate on production. This is far from how the world works at the moment though, major changes in the way we think about this is required.
The other model that makes sense is for all this content to be torrent based - this automatically distributes the bandwidth load where it is being used (assuming people don't leach too much! maintain a ratio etc.) This requires an even more radical rethink of how the content is thought of currently...
In short, there is, with it's current centralized distribution model, reason for bodies like the BBC to geographically restrict content. You need to de-centralise the distribution model for it to be fair otherwise. People don't think of it all like that right now though.
Hmm, that's kind of piddle and crap compare to what I was looking at today, one department have over a million folders (24 terabytes) don't know how many files yet: taking too long. I'd be suspicious of performance scaling well to that sort of level.
sorry, the user friendly bit was completely a mod edit, I asked for redundant or highly available or something: not user friendly.
Sounds cool, seems to not have been properly implemented yet? Still in development?
Highly available? How do you do that? I don't want remote backups, I want a highly available system that can just start using a second node transparently when one goes down. DFS/VSS does that for you.
ZFS looks great for snapshots... does it do anything in a distributed manner? That is, can it's storage pools be on different physical systems? how can you do ZFS without a single point of failure? is it possible?
Do any of these options work without a single point of failure? Ceph by itself sounds awesome, but how do you get versioning of some sort on top of it without introducing a single point of failure? I guess you could stick load balances in front of things like the WebDAV option, but that is getting very complicated. DFS/VSS does quite easily introduce a system without a single point of failure and is still looking like it's mighty useful compared to all available in the Linux world.
I think you present the first real options of all the posts here but they are not easy options.
What is the performance of this sort of thing like? I'm thinking of the case of a business using this constantly. Is it going to work for a few hundred users using a share daily? Or is that just going to make it die? I was thinking something that was a copy-on-write type thing, not a slow cron job type thing, but if it does perform OK...
Well, to be more specific the conductor facilitates a single interpretation in the change of pulse and other variables open to interpretation. A good orchestra can easily keep a steady pulse and play together rhythmically without a conductor just fine. They can even start together blindfolded, this is about listening to each other, esp breathing, it is actually not as hard as you might think. A very good orchestra can even come to a good consensus as to musical interpretation without a conductor, but will generally come together much faster with a good conductor. A very good orchestra will completely ignore a bad conductor during a performance and sound better for it.
School orchestras and the like will have conductors perform a more "keep everyone together rhythmically" type function, but this is not the ideal situation. More of an aid to learning situation.
An independent! Parties suck because they are parties.
The greens are not so bad. They used to be radical hippies but they grew up, they would be my next choice after an independent. They do suffer from that past image still, but their policies now look the most sensible of any party.
Unless they are trying to find a free channel for data and there just isn't one because 6 people around you are doing this all on different channels. Allocation of another part of the spectrum for power transmission: much better idea than trying to use the same ones as for data.
Using incidental energy that is being used for data transmission anyway - nothing wrong with that. Flooding data channels with noise, not so nice for other people around you.
"No Party may require a service supplier, as a condition for supplying a service or investing in its territory, to: (a) use computing facilities located in the Party’s territory."
So my reasoning for not using your USA located computing facilities is not because the are in the USA, it is because you can not grantee the level of data security I require at that facility. The fact that this happens to be because the facility is located in a particular territory with stupid laws - relevant but NOT the end reason I'm refusing to deal with it. The reason is security requirements I have, not physical location. Not a problem?
I think you are trying to argue about different issues thinking they are the same.
ShanghaiBill is talking about the law being contrary to the way he thinks is more effective.
gstoddart is talking about Uber not complying to the laws as they are.
Both point are relevant to the article, but not the same.
More info on the specs of the "supercomputer" that TFA only glossed over:
The result is the custom-built supercomputer, which we call
Minwa . It is comprised of 36 server nodes, each with 2
six-core Intel Xeon E5-2620 processors. Each sever con-
tains 4 Nvidia Tesla K40m GPUs and one FDR InfiniBand
(56Gb/s) which is a high-performance low-latency inter-
connection and supports RDMA. The peak single precision
floating point performance of each GPU is 4.29TFlops and
each GPU has 12GB of memory. Thanks to the GPUDirect
RDMA, the InfiniBand network interface can access the re-
mote GPU memory without involvement from the CPU. All
the server nodes are connected to the InfiniBand switch.
Figure 1 shows the system architecture. The system runs
Linux with CUDA 6.0 and MPI MVAPICH2, which also
enables GPUDirect RDMA.
In total, Minwa has 6.9TB host memory, 1.7TB device
memory, and about 0.6PFlops theoretical single precision peak performance.
It's not that powerful overall, but seems to be well thought out for what it is doing. I do the see point about fudging data somehow, they do provide a lot of information of what they supposedly did here
I don't know how this is verifiable, it's not like they have released source code or binaries for the software as far as I can tell.
Australia fixes most of the problems you describe simply: voting is mandatory, you must give time for employees to go and vote or face very stiff fines.
ARIA produces some "top charts" in Australia, here's how.
This is much better than what they used to do many years ago: rely on distribution figures (that is, how many unit were shipped this week.) I still don't think they reflect much except the skills of advertisers.
That is all true, but the economic implications for some situations make trading off privacy and control worth while. If you are not a tech savvy individual or company large enough to make the economy of scale worth while for your own IT department SAS can be a good thing economically.
Think about the situation of a small business with 10 employees. You want the ten employees to have company email and be able to share documents with each other. Do you:
1) buy a server, a backup solution, buy all the software required, engage a consulting company to get it all going for you and say support to said company when stuff breaks
2) buy a server, a backup solution, buy all the software required, hire an IT guy part time forever (assuming you can find a good one who will work part time)
3)pay $5 a month per user (Australian price, not sure about elsewhere) for google apps and gmail and be done with it.
Yes, option three trades away control to a large extent and privacy to some extent, but does the small business care? Unlikely they do. Should they care? Another question all together, maybe they should some times. The convenience and economic benefits (i.e. cheap) are too good to pass over in this small business situation. Should a large company of 20,000 employees do the same? I don't think so, having your own infrastructure at that size makes a lot more sense.
As to the original question, I can see a place for both thick and think clients in the future. You have to think about who your end client is going to be and a product that suits one sort of client. One size does not fit all and one product is not going to dominate all of a market because it is thick or thin client. I don't think there will ever be a situation in the future where all, say, word-processors are either thick or thin clients because this is always what is best.