My total monthly electric consumption these days is well under 100 KWh.
My god, what kind of house do you have to use that much electricity, and where do you live? My house doesn't use a quarter of that in a full year, including heating. And I'm in Canada!
Here we have 160 users for about 13GB of emails. We do ont have limits, not do we expect to put any.
Email is the lifeblood of the company, and it is handled with the respect it deserves. 13GB is actualy a very small amount of data compared to all the other stuff we handle.
Setup: Solaris + SAN + Postfix + Dovecot (IMAP)
Most users have all their emails (in and out) ever since they started here.
www.flextps.org is a GPL package that works really well with Axis video servers. Its main purpose is to stream video streams over the web, but it also has a DVR functionality where you specify which streams you want to record, the frame rate and the duration of recording. It's all perl-based and you could probably use a cronjob to start a 24h recording every midnight.
To use you airline example - deregulation has saved consumers tons of money. Fare staht used to be $3000 or $1500 are now $300 or $150. Airlines that built big networks when they were assurred of a profit through regulation are struggling; but once one or two major airlines fail then the remaining onse will make money. Unfortunately, our government can't seem to let them fail which just reinforces their refusal to change. Of ocurse, politicians realize teh marginal second tier cities that today get jet service will lose it once airlines price fsares at realistic costs and we can't have Podunk ID lossing its 2x a day jet service - think of the children!! So airlines get bailouts so politicians districts keep air service.
Please, learn to type! You *may* have good points here, but they're lost when someone needs a decoder to understand what you type.
I'm a consultant - I convert gibberish into cash-flow.
A few years ago I applied to get a Quebec Birth Certificate. I had to apply in person, but the questions they asked were simple enough for an identity thief who's well prepared, but hard enough for me: they needed to know the name (in addition to the standard name and birth date) of the hospital where I was born (what about hose born at home ?) and the name of the church I was baptised in. Of course I remembered neither so had to call my mom...:-)
..except in Quebec (just to make it confusing), where everyone goes to college (called CEGEP) after high school
Of course, High school in Quebec is the equivalent of middle school in the US, and CEGEP is the equivalent of high school in the USA. Some CEGEP offer a two year program to get you ready for university, while others offer a 3 to 5 year professional training program... Confused yet?
Since when did the USA decide North and South America are different continents? Pretty much everyone else in America seems to think otherwise. And I'm referring to the continent, by the way.
Amen to that. That's what I learned in school: America is a continent, USA is a country. I just checked in the dictionary: My French Larousse 1996 agrees, America is a continent made up of three regions (north, central, south). However, the Encylopedia Britannica doesn't agree, and list South America and North America as separate continents, with no indication on where's the separation.
When I switched to an Apple laptop, and made my company add Apple Desktops to our mostly Linux and Solaris environment, it's the operating system, not the hardware platform that interested us the most: Unix back-end, source compatibility for all our applications, and support for office applications.
Previously, we had people switch from machine to machine depending on which application they need. OS X has not removed the need for Windows, but it has much reduced it.
The reason why 'A' is a bad idea can be found in the paper 'End to End Arguments in System Design' [1].
But basically, adding an additional layer would increase latency for ALL packets (because of the additional checks), to optimize for a relatively rare occurence.
As for 'B', addresses are layer 3 while ports are layer 4. You can already decide not to use ports and do data transfers over raw IP (this is how ping is usually implemented).
However, the concept of port simplifies a lot the code of daemons. A daemon can bind on a specific port for all addresses to listen on all network interfaces of a server. Doing the same with your scheme would mean either:
- the application needs to listen on multiple sockets (more complicated); or
- the 'well known address' is applied to a virtual interface, the server therefore needs to internaly route packets and needs to advertise a route to its internal network - more work for the routers.
So, it'd be a lose-lose proposition.
I do work in academia, and one thing I've noticed is that more and more funding agencies (NSF among others) is that they are more likely to fund your proposal if the proposal states that the software created by the project will be released under an open source license. Open Source is good for the funding agency, as it means that the research will be more widely disseminated, and so will have a bigger bang for the buck. It can also be folded into the 'educational outreach' part of the proposal.
Once the funding agency has agreed to fund a proposal which stated that the software would be released under an open source license, this thing goes into the contract between the agency and the university, and you're in the clear.
The university would be really stupid to refuse such a contract - in fact, I've never heard of a university refusing money.:-)
Same here. Computers are configured to default to no screen saver, but the monitor goes in power save after 20 minutes idle. The hard disk goes to sleep after 3 hours of idle time. We don't want the computers to sleep completely, as we often use idle desktops as batch processing machines.
It's not just the fact that the skills required of a programmer and sysadmin are different, the dynamics are also vastly different. Programming is a long haul job, where you need to be able to focus on the job at hand. As a sysadmin, you're constantly being interupted by support requests (unless you happen to have a separate help desk). Since the sysadmin jobs are always urgent (email doesn't work, web server down, file server acting up, network is slow, etc), the programming job keeps taking a back seat.
Second this. We have a meeting room with a Polycom VS4000 connected to a 42" plasma on one side, and a SMARTboard on the other. The SMARTboard is used to run applications and make annotations. The polycom will send either the SMARTBoard image or the camera image (based on users' wish). The conference camera is a sony PZT. Everything is controlled through an AMX touch screen (plasma, projectors, image source for the polycom, speakers volume, lights, etc).
The polycom is only connected to the 'net, we don't use ISDN for conference, though the unit is able to.
All the equipment is in a a rack in a locked room, the only interface is the AMX touch screen, which makes it very user friendly.
In summary, Microsoft provided the ability to make the system more secure using non-privileged accounts and groups like every other major OS, but application developers are not taking advantage of it
You do realize that Microsoft, if they were serious about security, could have fixed that with the release of Windows XP. For some reason, most application publishers want the 'designed for Windows XP' sticker, logo or whatever. To get this, they're supposed to follow the guidelines of the program setup by Microsoft. For some unknown reason, Microsoft has never asked that the programs be well-behaved in multi-user, non-administrator environment. So developpers don't care (path of least resistance).
In most cases, fixing the issues are simply to store preference files in the right place (user's directory, user's registry).
Furthermore, the buffer overflows in quicktime do not afford an attacker root priviledges, do they?
Correct. On OS X, users do not run with root privilege. Running as root requires special steps, as it isn't even offered as an option in the graphical login window. Configuring a user as an administrator in OS X only means that the user can become root (through sudo) after typing his password for the duration of a single task.
By comparison, new users created in Windows have full administrator privileges. Even the filesystem permissions are so wide open that a restricted user is able to delete critical files. This is why most Windows machines, even when properly patched, are much more vulnerable to worms/trojan horses than unix-based machines.
NEES seems to be poised to become the focus of federaly-funded earthquake research in the US. Part of the deal is that NEES provides a repository for research data that will survive individual grants, and requires published research to be made available in the repository.
I say GO to such efforts. The more information is out there, the better.
Except that management will often count the tech guy's time as something that's already paid for.
That depends. Many companies, especially bigger ones, do per-project accounting to determine which are profitable and that usually include all manpower costs, even that of the admins whose salary has to be paid anyway.
As for the quarterly budget, the admin having more free time means he can work on yet another project, instead of having to hire a second admin, that has a sizeable impact.
So it all boils down to the ratio of admins vs computers. A single admin per 10 machines can afford to build his own, but if you're administrating 200 servers, you definitely can't.
wouldn't you put nearly the same effort researching your server hardware as well?
No, that's the whole point of buying from a well-known manufacturer. And Dell aren't that great either, at least compared to IBM, HP, Sun...
My day job is administrating hundreds of servers, and I wouldn't get half as much done if I had to build and qualify my own servers, and if I couldn't rely on several years of manufacturer tech support. Did you know that with IBM's standard warranty on servers a tech comes out and replaces the parts for you, if you want? That includes troubleshooting when a server doesn't behave but the cause can't be directly determined.
I've never had RAM fail in any capacity, so I can't say spending there will net great rewards
I have, and having the system say memory error corrected is so much better than random lockups and faulty operation. Get ECC memory if you value reliability and correctness.
The problem I've found is that while it's possible to setup a workstation as a server and get good performance and reliability, it's so much work to research and build that it's often more cost-effective to just buy server grade hardware. When you start counting your cost as part of the system machine, a server no longer seem expensive.
The Keyspan Digital Media Remote is a small IR controller. It comes with a base that connects to the PC or Mac using USB. It has keys for play, stop, rewind, forward, next, previous, etc. I've tested it with PowerPoint and various media programs, and it works at least as far away as 25'. And cheap to boot, I got mine for $35 at the local computer store.
More info.http://www.keyspan.com/products/usb/remote/
The university I work has quite a number of $100K+ clusters that run various versions of free operating systems, and are usually maintained by grad students. The priority is to get as many CPU and as much memory as possible for a given amount of money. We then build it ourself as student time is comparatively cheap.
I agree with this post. I was a Speakeasy customer for a while until the DSL provider decider to cut service to my CO. Now I'm with acmenet.net, a local provider.
Both providers were roughly the same: $45 per month for SDSL 384Kbps (symmetric) with 3 static IP addresses and no filtering. Pretty good tech support, too.
My total monthly electric consumption these days is well under 100 KWh.
My god, what kind of house do you have to use that much electricity, and where do you live? My house doesn't use a quarter of that in a full year, including heating. And I'm in Canada!
Here we have 160 users for about 13GB of emails. We do ont have limits, not do we expect to put any. Email is the lifeblood of the company, and it is handled with the respect it deserves. 13GB is actualy a very small amount of data compared to all the other stuff we handle. Setup: Solaris + SAN + Postfix + Dovecot (IMAP) Most users have all their emails (in and out) ever since they started here.
www.flextps.org is a GPL package that works really well with Axis video servers. Its main purpose is to stream video streams over the web, but it also has a DVR functionality where you specify which streams you want to record, the frame rate and the duration of recording. It's all perl-based and you could probably use a cronjob to start a 24h recording every midnight.
Please, learn to type! You *may* have good points here, but they're lost when someone needs a decoder to understand what you type.
I'm a consultant - I convert gibberish into cash-flow.
From your text above, you produce gibberish.
A few years ago I applied to get a Quebec Birth Certificate. I had to apply in person, but the questions they asked were simple enough for an identity thief who's well prepared, but hard enough for me: they needed to know the name (in addition to the standard name and birth date) of the hospital where I was born (what about hose born at home ?) and the name of the church I was baptised in. Of course I remembered neither so had to call my mom... :-)
(Taking the bait...)
So, what's the coolest function of VLAN, other than creating two broadcast domains on the same switch?
Of course, High school in Quebec is the equivalent of middle school in the US, and CEGEP is the equivalent of high school in the USA. Some CEGEP offer a two year program to get you ready for university, while others offer a 3 to 5 year professional training program... Confused yet?
Amen to that. That's what I learned in school: America is a continent, USA is a country. I just checked in the dictionary: My French Larousse 1996 agrees, America is a continent made up of three regions (north, central, south). However, the Encylopedia Britannica doesn't agree, and list South America and North America as separate continents, with no indication on where's the separation.
When I switched to an Apple laptop, and made my company add Apple Desktops to our mostly Linux and Solaris environment, it's the operating system, not the hardware platform that interested us the most: Unix back-end, source compatibility for all our applications, and support for office applications. Previously, we had people switch from machine to machine depending on which application they need. OS X has not removed the need for Windows, but it has much reduced it.
The reason why 'A' is a bad idea can be found in the paper 'End to End Arguments in System Design' [1]. But basically, adding an additional layer would increase latency for ALL packets (because of the additional checks), to optimize for a relatively rare occurence. As for 'B', addresses are layer 3 while ports are layer 4. You can already decide not to use ports and do data transfers over raw IP (this is how ping is usually implemented). However, the concept of port simplifies a lot the code of daemons. A daemon can bind on a specific port for all addresses to listen on all network interfaces of a server. Doing the same with your scheme would mean either: - the application needs to listen on multiple sockets (more complicated); or - the 'well known address' is applied to a virtual interface, the server therefore needs to internaly route packets and needs to advertise a route to its internal network - more work for the routers. So, it'd be a lose-lose proposition.
Once the funding agency has agreed to fund a proposal which stated that the software would be released under an open source license, this thing goes into the contract between the agency and the university, and you're in the clear.
The university would be really stupid to refuse such a contract - in fact, I've never heard of a university refusing money. :-)
Same here. Computers are configured to default to no screen saver, but the monitor goes in power save after 20 minutes idle. The hard disk goes to sleep after 3 hours of idle time. We don't want the computers to sleep completely, as we often use idle desktops as batch processing machines.
It's not just the fact that the skills required of a programmer and sysadmin are different, the dynamics are also vastly different. Programming is a long haul job, where you need to be able to focus on the job at hand. As a sysadmin, you're constantly being interupted by support requests (unless you happen to have a separate help desk). Since the sysadmin jobs are always urgent (email doesn't work, web server down, file server acting up, network is slow, etc), the programming job keeps taking a back seat.
Second this.
We have a meeting room with a Polycom VS4000 connected to a 42" plasma on one side, and a SMARTboard on the other. The SMARTboard is used to run applications and make annotations. The polycom will send either the SMARTBoard image or the camera image (based on users' wish). The conference camera is a sony PZT. Everything is controlled through an AMX touch screen (plasma, projectors, image source for the polycom, speakers volume, lights, etc).
The polycom is only connected to the 'net, we don't use ISDN for conference, though the unit is able to.
All the equipment is in a a rack in a locked room, the only interface is the AMX touch screen, which makes it very user friendly.
You do realize that Microsoft, if they were serious about security, could have fixed that with the release of Windows XP. For some reason, most application publishers want the 'designed for Windows XP' sticker, logo or whatever. To get this, they're supposed to follow the guidelines of the program setup by Microsoft. For some unknown reason, Microsoft has never asked that the programs be well-behaved in multi-user, non-administrator environment. So developpers don't care (path of least resistance).
In most cases, fixing the issues are simply to store preference files in the right place (user's directory, user's registry).
Correct. On OS X, users do not run with root privilege. Running as root requires special steps, as it isn't even offered as an option in the graphical login window. Configuring a user as an administrator in OS X only means that the user can become root (through sudo) after typing his password for the duration of a single task.
By comparison, new users created in Windows have full administrator privileges. Even the filesystem permissions are so wide open that a restricted user is able to delete critical files. This is why most Windows machines, even when properly patched, are much more vulnerable to worms/trojan horses than unix-based machines.
I say GO to such efforts. The more information is out there, the better.
That depends. Many companies, especially bigger ones, do per-project accounting to determine which are profitable and that usually include all manpower costs, even that of the admins whose salary has to be paid anyway.
As for the quarterly budget, the admin having more free time means he can work on yet another project, instead of having to hire a second admin, that has a sizeable impact. So it all boils down to the ratio of admins vs computers. A single admin per 10 machines can afford to build his own, but if you're administrating 200 servers, you definitely can't.
No, that's the whole point of buying from a well-known manufacturer. And Dell aren't that great either, at least compared to IBM, HP, Sun...
My day job is administrating hundreds of servers, and I wouldn't get half as much done if I had to build and qualify my own servers, and if I couldn't rely on several years of manufacturer tech support. Did you know that with IBM's standard warranty on servers a tech comes out and replaces the parts for you, if you want? That includes troubleshooting when a server doesn't behave but the cause can't be directly determined.
I have, and having the system say memory error corrected is so much better than random lockups and faulty operation. Get ECC memory if you value reliability and correctness.
The problem I've found is that while it's possible to setup a workstation as a server and get good performance and reliability, it's so much work to research and build that it's often more cost-effective to just buy server grade hardware. When you start counting your cost as part of the system machine, a server no longer seem expensive.
You're talking about the PIX - Cisco's firewall product line. Those are software only, and if you pirate the software you can easily clone it.
But the real iron does everything in hardware (except error handling).
The Keyspan Digital Media Remote is a small IR controller. It comes with a base that connects to the PC or Mac using USB. It has keys for play, stop, rewind, forward, next, previous, etc. I've tested it with PowerPoint and various media programs, and it works at least as far away as 25'. And cheap to boot, I got mine for $35 at the local computer store. More info.http://www.keyspan.com/products/usb/remote/
The university I work has quite a number of $100K+ clusters that run various versions of free operating systems, and are usually maintained by grad students. The priority is to get as many CPU and as much memory as possible for a given amount of money. We then build it ourself as student time is comparatively cheap.
I agree with this post. I was a Speakeasy customer for a while until the DSL provider decider to cut service to my CO. Now I'm with acmenet.net, a local provider.
Both providers were roughly the same: $45 per month for SDSL 384Kbps (symmetric) with 3 static IP addresses and no filtering. Pretty good tech support, too.
Soooo... Instead of a single point of failure, you would want to have four single point of failure ? Makes tons of sense, of course.