Well, I don't know about you, but the university I'm affiliated with is in talks with RedHat right now about that issue. The deal is this, for ~3k a year we will license the RedHat proxy server. Which, more or less, is a private mirror of their distribution server (ES, AS, WS, and errata updates). Students, faculty, and staff will have free access to the server to install WS on all their personal platforms.
Any university owned equipment (lab workstations, servers, etc.) will also feed off the central proxy/distribution server but will operate on a separately negotiated license.
So for you as a student, this will give you a free copy of WS to put on any number of computers (non university owned) you wish and entitles you to all the updates while you attend. It even gets you access to a private mirror so you don't have to contend with others on the public servers.
As for uni wide costs, this is chump change. The savings in labor, expertise, and steep discount being offered on the university owned equipment licenses make this very attractive. We know that two years from now, our installed base of workstations and servers will still be able to easily get its hands on the latest SSL/SSH/errata of the month.
When we think of support, we think of the 1000s of lab workstations and servers. We also have to keep in mind the folks who keep them patched and up to date. Which, in a lot of circumstances, are student employees who will be gone in a year or so. It saves for us to implement a patch process that is easy to understand and has guarantees attached to it.
We pay Sun and MS the same for the same ability on their products. We can't afford to have some guru create an install base off a distribution that we can't be assured will offer timely errata updates and/or upgrate paths. We gladely pay to have that capability. And in the process give back to the community through RedHat's contributions to Linux.
Gentoo, Debian, and White Box are all fine distributions. But we can't assume the risk of having to stand in front of a board explaining how a 1000 computers got rooted and started pumping out spam to the world. Especially because we chose a distribution that lost funding or interest and couldn't supply errate updates or easy upgrade paths.
FYI, I'm just hoping to give a perspective from the other side of the fence. We are RedHat's target market and are investigating it just like we've investigated other avenues.
Just to clarify. By printed ballot, I'm not talking about a cash register knock off receipt printer installed in the machine that accumulates votes on a role of paper purchased at the local Office Max. I'm thinking of something that will last at least until the end of a recount and/or judicial judgement without ink fade. It should be easily human readable and have something like MICR characters on it for machine readability.
While I'm at it, I would also like to request the specs on the format so that I may present myself as a third party validator if needed...
Anyhow, it's late and I'm going to bed. Tomorrow, I'll awake back into the current reality...
I may be paranoid here of just ignorant of how the final process works, but my biggest fear from these things involves the transfer of the data from the precinct machines back to its central point.
After the machines are dropped off and loaded into the "rented truck" some tech has to extract and accumlate the count data. How easy would it be to alter the counts before the data is transmitted to the central site? Suppose a non tech centric deputy was to oversee the final tally. It might be possible for the tech to alter the counts in plain sight of the deputy without him knowing specifically what's going on.
It's really hard for me to believe with the amount of tin foil programming/tech talent available that these systems have gone into use without something as simple as a printed ballot. There too much black box magic going on in these things for me to trust them...
Actually, it wouldn't suprise me if this is correct. If you notice, he was reading the 500MB file while a continuous streaming write was going on in the background. On 2.4.x, a write streamout will kill read performance drastically. Mostly due to the way the I/O scheduler schedules the read. Which, most of the time, is to stash it at the end of the writes.
The two new I/O schedulers in 2.6.x help to resolve this. For more info, check here.
It is my understanding that you are not compelled to keep logs under the PATRIOT act. What is required is that you have the ability to trace/tap/monitor specific activities at any time.
So a scenario could be this. FBI walks in and puts down a peice of paper that states they need a trace of information on user XYZ or user currently at IP address 1.2.3.4. You are not required to deliver any information at the time the request is served. What you are required to do is to implement a trace on logs/email/etc for data related to the request. The FBI is then required to deliver a warrant within 60 days to retrieve the requested information. If the warrant isn't delivered withing 60 days, you can halt the tap/trace/whatever and dispose of the data.
IANAL and the above is just my understanding of the issue.
A perfect solution would be a form of network block device that mounts distributed NBD shares. The Linux DRBD Project has this capability. From their website, "You could see it as a network raid-1".
Re:Finding Bin Laden with mini sensor surveillance
on
The Smart Sensor Web
·
· Score: 1
Still pretty popular. This project is to create a bunch of little sensors that folks can scatter about. They'll then form a sort of redundant mesh. Don't know much about it, their posters on the wall look pretty neat:).
More Info
I'm assuming the mesh network all runs on the same channel? How well does this thing scale if, for example, I cover a 1 square mile urban area with 10 nodes and only one of those nodes has a wired connection? I would think that the channel would become saturated rather quickly.
In addition, IBM plans to offer the processor as the centerpiece of future Linux-based systems
Hopefully IBM will offer reference mobo designs for the new GPUL line. This may spur alternate price conscious server alternatives (along with x86-64). Especially if Intel is still dragging their ass...
The statistic I heard is that already 20% of all long distance calls are on VOIP.
I suppose there would be incredible incentive for a company to setup a VoIP infrastructure using standard leased links carrying IP traffic. I'm sure the cost savings would be very substantial. I know of some call centers that trunk calls using VoIP to their satelite centers over leased private networks.
I also don't doubt that VoIP traffice over the internet is rising in the single digits per year. Fairly insignificant if the true non VoIP traffic is doubling every year. But, would the sheer number of cell phones in use cause this statistic to explode beyond belief? Perhaps to the point where VoIP traffic is on par with the data traffic.
If the above theory would become a reality, I can imagine wars breaking out amongst the backbone providers over who's going to absorb the costs. Say backbone provider B is a transient between providers A and C. Furthermore, say a healthy percentage of B's traffic is VoIP traffic originating from A to C or C to A. Provider B may get a little discouraged over the idea that the amount of transient traffic (peer traffic) vs. the amount of originated traffic (you know, the bandwitdh they actually charge for) is climbing steadily.
It would be interesting to see some numbers regarding the number of cell phones in use vs. the number of allocated IP addresses within the U.S. Almost everyone I know of has a cell phone.
Of course, I also agree that this is futher off than three years. As a previous poster already stated, the latencies your see over the net, most of the time, aren't ideal for voice.
It seems today that everyone is trying to take something designed to operate over a private/proprietary network and push it out over the internet. In the article, Waryas says, "there is a potential return for carriers using the IP backbone to transfer calls versus digital signals". If this were to materialize over the next three years, what can we expect of the internet backbone as a whole? Phones already have an addressing system (aka a "phone number"), is it a good idea to provide an IP address to _every_ cell phone? Perhaps IPv6 could help here, but it also begs the question of who will absorb the added bandwith utilization/costs? A call from east coast to west coast could traverse two or three backbone providers. I can just imagine the congestion at the peer points.
Don't get me wrong, it would be really nice to have a handheld device or laptop + handheld/cell phone to be able to natively handle IP traffic, but I also wonder what the long term affects would be on the traditional internet...
Heh, I also work for a BIG 10 University. I don't do research, but work on the operations side of things. I can say, our budget got cut almost 10% this year. Though, they aren't laying off researchers or faculty, they are laying off lower end staff (think, cafeteria workers, custodial workers, etc).
The catch phrase that we go by is, "Those who can't make it in the private sector, wind up in the public sector."
Every one at the management level is too preoccupied with:
1> Increasing their ego. 2> Stiffling others to ensure their position's safety. 3> Wasting their staff's time, primarily because they're not qualified for their position.
Job safety is somewhat of a bonus. Believe me though, if you step on the wrong toes, they'll riff your ass to a lower position where you have absolutely no authority. Some of the most innovative people that I work with wound up in the lower bowels because their innovative ideas where viewed as threatening by others.
All in all, working in academia has produced three times the amount of stress that I used to deal with in the private sector. The main difference now is the fact that I get paid 10k less per year than I did in the private sector.
Overall, my experiences in academia have been very disappointing. I thought folks of higher intelligence would be more prevalent, I thought wrong.
I happen to work for one of the largest universities in the U.S. and we purchase most of our servers from Sun. Educational discount for us is typically 40%.
I'm going out on a ledge, but I'd believe one of the major applications of this is in DWDM down single mode fiber. It simplifies manufacturing costs to have a laser that can output on different wavelengths vs. multiple lasers all specifically designed for a single wavelength. I was reading a telecom book today at Barnes and Noble that pointed to that specifically as one reason for the current high costs of DWDM. Of course I have to take the book for a grain of salt considering it was yellow and black:).
Their marketdroids have a bad habit of rounding the values down and evening them off. This allows them to post bigger numbers on the actual size of the drive since dividing by 1000000 instead of 1048576 yeilds a larger end result.
Mod this guy up. Most folk in the application programming enviroment have never done industrial control programming before. When these guys are talking I/O, they're talking interfaces to Omron or Allen Bradley PLC controllers, Radio telemetry links, MODBUS acquired devices, and maybe some proprietary data acquisition stuff. Heh, and throw in the Opto-22 devices as well (probably over a 485 multi-drop link:).
I used to work at a large mortgage company here in the U.S. About a year ago, they scrapped a customer workbench application that was (to that development point) written in java. The reason being is that the cost of replacing every agent PC (over a 1000 of them) with faster processors and more RAM was too cost prohibiting. The agent PCs at the time were midrange PII's (333's and up) with 64MB of ram. The decision was made after doing numerous tests with the agents testing the application and recording their response to it. Agents were frustrated with the lack of response from the application (which is vital when your on the phone with a customer). From the test, they determined that to attain the performance required, they'd have to retrofit the agent PCs to at least PIII-650 with 256MB of ram. This workbench app is a very large application. From that point, development switched to VC++ to keep the vested cost they already had in their agent setups.
Slashdot Mirror
Well, I don't know about you, but the university I'm affiliated with is in talks with RedHat right now about that issue. The deal is this, for ~3k a year we will license the RedHat proxy server. Which, more or less, is a private mirror of their distribution server (ES, AS, WS, and errata updates). Students, faculty, and staff will have free access to the server to install WS on all their personal platforms.
Any university owned equipment (lab workstations, servers, etc.) will also feed off the central proxy/distribution server but will operate on a separately negotiated license.
So for you as a student, this will give you a free copy of WS to put on any number of computers (non university owned) you wish and entitles you to all the updates while you attend. It even gets you access to a private mirror so you don't have to contend with others on the public servers.
As for uni wide costs, this is chump change. The savings in labor, expertise, and steep discount being offered on the university owned equipment licenses make this very attractive. We know that two years from now, our installed base of workstations and servers will still be able to easily get its hands on the latest SSL/SSH/errata of the month.
When we think of support, we think of the 1000s of lab workstations and servers. We also have to keep in mind the folks who keep them patched and up to date. Which, in a lot of circumstances, are student employees who will be gone in a year or so. It saves for us to implement a patch process that is easy to understand and has guarantees attached to it.
We pay Sun and MS the same for the same ability on their products. We can't afford to have some guru create an install base off a distribution that we can't be assured will offer timely errata updates and/or upgrate paths. We gladely pay to have that capability. And in the process give back to the community through RedHat's contributions to Linux.
Gentoo, Debian, and White Box are all fine distributions. But we can't assume the risk of having to stand in front of a board explaining how a 1000 computers got rooted and started pumping out spam to the world. Especially because we chose a distribution that lost funding or interest and couldn't supply errate updates or easy upgrade paths.
FYI, I'm just hoping to give a perspective from the other side of the fence. We are RedHat's target market and are investigating it just like we've investigated other avenues.
Guess what else Paul Allen funds .
Just to clarify. By printed ballot, I'm not talking about a cash register knock off receipt printer installed in the machine that accumulates votes on a role of paper purchased at the local Office Max. I'm thinking of something that will last at least until the end of a recount and/or judicial judgement without ink fade. It should be easily human readable and have something like MICR characters on it for machine readability.
While I'm at it, I would also like to request the specs on the format so that I may present myself as a third party validator if needed...
Anyhow, it's late and I'm going to bed. Tomorrow, I'll awake back into the current reality...
I may be paranoid here of just ignorant of how the final process works, but my biggest fear from these things involves the transfer of the data from the precinct machines back to its central point.
After the machines are dropped off and loaded into the "rented truck" some tech has to extract and accumlate the count data. How easy would it be to alter the counts before the data is transmitted to the central site? Suppose a non tech centric deputy was to oversee the final tally. It might be possible for the tech to alter the counts in plain sight of the deputy without him knowing specifically what's going on.
It's really hard for me to believe with the amount of tin foil programming/tech talent available that these systems have gone into use without something as simple as a printed ballot. There too much black box magic going on in these things for me to trust them...
If that's the intended use. Perhaps they could team up with NASA which already has a winged flying prototype.
Keep in mind when DARPA talks about adhoc networks and such, they also have stuff like this in mind...
Actually, it wouldn't suprise me if this is correct. If you notice, he was reading the 500MB file while a continuous streaming write was going on in the background. On 2.4.x, a write streamout will kill read performance drastically. Mostly due to the way the I/O scheduler schedules the read. Which, most of the time, is to stash it at the end of the writes.
The two new I/O schedulers in 2.6.x help to resolve this. For more info, check here.
Put this in /etc/sysctl.conf
:).
kernel.panic = 120
That will tell the kernel to reboot itself 2 minutes after a panic. It has saved me in the past before
It is my understanding that you are not compelled to keep logs under the PATRIOT act. What is required is that you have the ability to trace/tap/monitor specific activities at any time.
So a scenario could be this. FBI walks in and puts down a peice of paper that states they need a trace of information on user XYZ or user currently at IP address 1.2.3.4. You are not required to deliver any information at the time the request is served. What you are required to do is to implement a trace on logs/email/etc for data related to the request. The FBI is then required to deliver a warrant within 60 days to retrieve the requested information. If the warrant isn't delivered withing 60 days, you can halt the tap/trace/whatever and dispose of the data.
IANAL and the above is just my understanding of the issue.
Not that anybody doesn't trust the parent poster, but sig the updates to ClamAV can be verified here.
A perfect solution would be a form of network block device that mounts distributed NBD shares. The Linux DRBD Project has this capability. From their website, "You could see it as a network raid-1".
Still pretty popular. This project is to create a bunch of little sensors that folks can scatter about. They'll then form a sort of redundant mesh. Don't know much about it, their posters on the wall look pretty neat :).
More Info
I'm assuming the mesh network all runs on the same channel? How well does this thing scale if, for example, I cover a 1 square mile urban area with 10 nodes and only one of those nodes has a wired connection? I would think that the channel would become saturated rather quickly.
They're really paying for is the true origin of their penile size. Is it really a product of nature or nurture...
In addition, IBM plans to offer the processor as the centerpiece of future Linux-based systems
Hopefully IBM will offer reference mobo designs for the new GPUL line. This may spur alternate price conscious server alternatives (along with x86-64). Especially if Intel is still dragging their ass...
The statistic I heard is that already 20% of all long distance calls are on VOIP.
I suppose there would be incredible incentive for a company to setup a VoIP infrastructure using standard leased links carrying IP traffic. I'm sure the cost savings would be very substantial. I know of some call centers that trunk calls using VoIP to their satelite centers over leased private networks.
I also don't doubt that VoIP traffice over the internet is rising in the single digits per year. Fairly insignificant if the true non VoIP traffic is doubling every year. But, would the sheer number of cell phones in use cause this statistic to explode beyond belief? Perhaps to the point where VoIP traffic is on par with the data traffic.
If the above theory would become a reality, I can imagine wars breaking out amongst the backbone providers over who's going to absorb the costs. Say backbone provider B is a transient between providers A and C. Furthermore, say a healthy percentage of B's traffic is VoIP traffic originating from A to C or C to A. Provider B may get a little discouraged over the idea that the amount of transient traffic (peer traffic) vs. the amount of originated traffic (you know, the bandwitdh they actually charge for) is climbing steadily.
It would be interesting to see some numbers regarding the number of cell phones in use vs. the number of allocated IP addresses within the U.S. Almost everyone I know of has a cell phone.
Of course, I also agree that this is futher off than three years. As a previous poster already stated, the latencies your see over the net, most of the time, aren't ideal for voice.
It seems today that everyone is trying to take something designed to operate over a private/proprietary network and push it out over the internet. In the article, Waryas says, "there is a potential return for carriers using the IP backbone to transfer calls versus digital signals". If this were to materialize over the next three years, what can we expect of the internet backbone as a whole? Phones already have an addressing system (aka a "phone number"), is it a good idea to provide an IP address to _every_ cell phone? Perhaps IPv6 could help here, but it also begs the question of who will absorb the added bandwith utilization/costs? A call from east coast to west coast could traverse two or three backbone providers. I can just imagine the congestion at the peer points.
Don't get me wrong, it would be really nice to have a handheld device or laptop + handheld/cell phone to be able to natively handle IP traffic, but I also wonder what the long term affects would be on the traditional internet...
Heh, I also work for a BIG 10 University. I don't do research, but work on the operations side of things. I can say, our budget got cut almost 10% this year. Though, they aren't laying off researchers or faculty, they are laying off lower end staff (think, cafeteria workers, custodial workers, etc).
The catch phrase that we go by is, "Those who can't make it in the private sector, wind up in the public sector."
Every one at the management level is too preoccupied with:
1> Increasing their ego.
2> Stiffling others to ensure their position's safety.
3> Wasting their staff's time, primarily because they're not qualified for their position.
Job safety is somewhat of a bonus. Believe me though, if you step on the wrong toes, they'll riff your ass to a lower position where you have absolutely no authority. Some of the most innovative people that I work with wound up in the lower bowels because their innovative ideas where viewed as threatening by others.
All in all, working in academia has produced three times the amount of stress that I used to deal with in the private sector. The main difference now is the fact that I get paid 10k less per year than I did in the private sector.
Overall, my experiences in academia have been very disappointing. I thought folks of higher intelligence would be more prevalent, I thought wrong.
FYI
I happen to work for one of the largest universities in the U.S. and we purchase most of our servers from Sun. Educational discount for us is typically 40%.
GENUITY (NET-GNTY-199-92) GNTY-199-92
199.92.0.0 - 199.95.255.255
Double Click, Inc. (NETBLK-DOUBLECLICK3) DOUBLECLICK3
199.95.206.0 - 199.95.209.255
Cable & Wireless USA (NETBLK-CW-10BLK) CW-10BLK
208.128.0.0 - 208.175.255.255
Inflow (NETBLK-CW-208-169-16A) CW-208-169-16A
208.169.16.0 - 208.169.23.255
MessageMedia (NETBLK-NETBLK-INFLOW-MMEDIA) NETBLK-INFLOW-MMEDIA
208.169.22.0 - 208.169.23.255
I'm going out on a ledge, but I'd believe one of the major applications of this is in DWDM down single mode fiber. It simplifies manufacturing costs to have a laser that can output on different wavelengths vs. multiple lasers all specifically designed for a single wavelength. I was reading a telecom book today at Barnes and Noble that pointed to that specifically as one reason for the current high costs of DWDM. Of course I have to take the book for a grain of salt considering it was yellow and black :).
Yup, except that to a hard drive manufacturer:
1 Terabyte = 1000GB = 1000000 MB
Their marketdroids have a bad habit of rounding the values down and evening them off. This allows them to post bigger numbers on the actual size of the drive since dividing by 1000000 instead of 1048576 yeilds a larger end result.
survive a direct attact when the ditch witch I rented three hours ago cuts through all the fiber conduits down the street.
Mod this guy up. Most folk in the application programming enviroment have never done industrial control programming before. When these guys are talking I/O, they're talking interfaces to Omron or Allen Bradley PLC controllers, Radio telemetry links, MODBUS acquired devices, and maybe some proprietary data acquisition stuff. Heh, and throw in the Opto-22 devices as well (probably over a 485 multi-drop link :).
I used to work at a large mortgage company here in the U.S. About a year ago, they scrapped a customer workbench application that was (to that development point) written in java. The reason being is that the cost of replacing every agent PC (over a 1000 of them) with faster processors and more RAM was too cost prohibiting. The agent PCs at the time were midrange PII's (333's and up) with 64MB of ram. The decision was made after doing numerous tests with the agents testing the application and recording their response to it. Agents were frustrated with the lack of response from the application (which is vital when your on the phone with a customer). From the test, they determined that to attain the performance required, they'd have to retrofit the agent PCs to at least PIII-650 with 256MB of ram. This workbench app is a very large application. From that point, development switched to VC++ to keep the vested cost they already had in their agent setups.