You also need a system that will allow invalidation of votes, and an inability to prove how you voted after the election.
This way when your boss asks you to vote a certain way on the job you can go ahead and do so, knowing you can change your vote later. And the next day you can't prove you voted one way or another so the boss is none the wiser.
You think big corporations control congress now? Wait until they literally hold tens of thousands of actual votes, and the ability to pay people for their votes (come into this booth with an unused voting card and we'll give you $10!).
Your contrived example is BS. All of these/8's have been allocated for years. The organizations holding them have had a decade to implement best practices and move away from such large, unnecessary utilizations of address space.
You make a value statement that it is better to not "waste" space.
If you look at the homes on your street (if you live in the US) they are probably not numbered sequentially. They probably skip every 10 numbers at least. This is so that if somebody puts up a new house you don't need to renumber the entire street. Sure, it is a waste of numbers, but numbers don't cost anything. It is more of a waste to not design for expansion.
In the same way these corporations are being smart in using their address space heirarchically, leaving room for growth. They are using a resource that is very cheap and plentiful (address space on a/8 network) to avoid investing in complex address assignment schemes.
These companies would care no more about renumbering their networks than a town might care to renumber their street addresses, even if it were practical to do so. Numbers are free - why not use them?
Staying with IPv4 is like solving the Y2K crisis by going to 99 month years, 99 day months, 99 hour days, 99 minute hours, and 99 second minutes. Sure, it prolongs the agony, but it makes no sense and ultimately doesn't solve the problem. Why not just start rolling out IPv6 where it is easy to do so? Design OS'es so that they obtain IPv6 addresses automatically via DHCP if they are available and use them, design routers so they give them out automatically, and eventually as stuff is replaced everybody will be running on IPv4 and IPv6. Then you can simply stop using IPv4 and nobody will notice. There is no reason to make some grand proclaimation that on 31-DEC-2007 we'll suddenly stop using v4 and start using v6.
Often the best solutions to problems are non-central ones. Sure, it isn't efficient in terms of address-space usage, but when you have more addresses than electrons in the universe should we really be concerned with that? To me it sounds like a bunch of IT guys standing in ivory towers saying that 32-bit addresses ought to be enough for anyone...
Funny you mention it. While not sharing in the analog partyline sense, PBX's already do this very thing. How many extentions are there on your office phone system? 20, 50, 100? How many actual CO lines are there? 8, 20? It's very rare for a PBX to have one CO line for each extention. In fact, that's counter productive -- just put a damned POTS phone on each desk. (Telco's used to sell this as centrex service.)
And likewise a corporate branch office I work at has about 25-30 globally routable addresses, and yet only one line running to and from it. That doesn't change the fact that it has globally-routable addresses. The line goes back to another corporate office, and only accesses the internet at large via proxy servers.
And there's equally no reason why they should. Do you really need to ping the nerf darts on my desk?
No, but perhaps you might have a need to do so when you are't in the office?
One major annoyance of mine is the difficulty in accessing home computers remotely due to NAT. This is a result of the general mindset among ISPs that IP addresses are a resource to be hoarded, and that only broadcasters should really have static ones, and everybody else should just be happy to use their interweb-TV.
The power of the Internet is P2P. However, P2P cannot operate correctly over NAT - at least not if there is NAT at both ends. As soon as you add a central server to proxy connections it is no longer truly P2P.
My whole point is that it doesn't cost a whole lot to allow everybody to just have their IP addresses, so why not just do it? Why should there be some burden of proof that I really need those addresses, as if I should care whether anybody else thinks I really need them or not. It would be one thing if they cost money to manufacture, but we're talking about numbers - they all exist already, we just need to start using them...
Franky, I tend to take the opposite appraoch. I make a point not to check voicemail often, unless I know somebody was planning on calling me.
If somebody has a problem, they can write me an email. I still won't read it right away, but it will get a response. In their email they will have had an opportunity to spell out their problem clearly, so in my response they'll get an actual answer. Typically when people leave a voicemail they just ask you to get back to them so they can explain their problem. I will usually just send them an email asking them to explain their problem.
If there is some kind of system outage it will be apparent from contacts from the official support organization, or by a flurry of phone calls. However, EVERYBODY wants their question answered right away, and I exercise extreme discretion at answering the phone as a result.
My customers have always evalulated me positively, and I think that it is because I meet their real needs. If customer management says that priority 1 is getting a certain project done, they don't want me to drop whatever I'm doing every time a subordinate wants to ask a question about a lower priority project. I allow my customer management to set my priorities, and I conciously apply my efforts where they make the most difference. This way, the largest number of people within the organization benefit.
Simply answering the phone every time it rings is allowing any random person to dictate your schedule. That is not likely to lead to effective use of time.
I already know your internal IPs. They have the form 192.160.x.y, or maybe 10.x.y.z. No reason that somebody can't write a virus that infects one border machine and then scans away at your internal network.
Most major corporations have routable addresses inside their networks - they don't hurt, so why not have them? They put up a nice firewall on all their entrances, and then it is just as secure as NAT (which is still only moderately secure - I'm sure NAT didn't protect many from Sasser and such - it still just takes one person with a laptop).
In most of these companies the internal extensions on the PBX are also world-dialable. It is cheaper than having a horde of operators to redirect calls, and more convenient.
Plus, when you have a corporate merger everybody has compatible phone numbers and IP addresses.
You speak of everything having an IP address as "pie in the sky". The fact is that if ISPs just started assigning routing IPv6 in addition to IPv4 it would just be a matter of time before everything transitioned. There is no reason everything shouldn't have its own IP address.
I'm sure back in the day of party lines the concept of having your own phone number was "pie in the sky". In any office only 10% of the people are on the phone at any time - why not just assign 1 phone number to each 10 desks and ask them to share...
Ok, you're assuming your IP addresses will change in the first place.
If you are a company with 50 computers, then you're just getting your IP space from your ISP, and that will be the case. As long as you have DHCP you'll still be fine. And for those two office copiers that don't support it you'll only need 2 minutes to fix them.
Migrations are only a headache when you have thousands of computers. In that case, you probably don't need to use your ISPs address space, but instead you can just get your own/16 or something like that. If you switch ISPs you can just keep the same IPs since you own them.
Here is a problem with NAT - mergers. What happens when two companies merge that have private networks? You suddenly have thousands of networks with the same addresses to remap. If on the other hand each company had a/16 you just need a router between them and no extra setup.
In an ideal world, one machine = 1 IP, globally routable. Sure, we use tricks when this isn't easy to implement (you can't very well have home dialup users taking their IPs with them when they switch ISPs unless you want the top-level routers to have to track every individual IP on the planet). However, NAT is really just a hack and is rarely the best solution to a problem...
Ok, while clearly most of these companies don't need a/8, envision yourself as the CIO of said company.
Underling: The public just called and wants to trade us a/16 or two for our/8.
CIO: Doesn't sound bad on the surface we don't have nearly that many systems. What is the downside.
Underling: Well, we gave every worldwide site their own/12. They gave each building a/16, those buildings gave each floor a/24 or two. Plus, we have a bunch of/16s and/12s for various testing purposes and dedicated private networks. We'd have to reprogram every router in the company - we only have about 500 of those. Then we'd have to do all kinds of software testing, and a bunch of systems with static IPs would need new ones.
CIO: That doesn't sound too hard. Get started on it.
Underling: Uh, how do you want to parse the IPs out - with the/8 we had space to spare and could just parse out large blocks to site heads, who could delegate it as they saw fit. With a/16 we need central management of all addresses so that we don't waste too much space. We need some system for keeping track of our addresses now...
CIO: Uh, I wasn't too good at this stuff in school - I was better at the marketing classes and learning how to go to dinner with vendor sales reps. Go on...
Underling: No problem, we just need a new web-based application that will use a database to...
Secretary: Excuse me, CIO, your golf appointment with the CEO is in 5 minutes.
CIO: Oh, I gotta run. How much money will we make by completely revaming our whole network?
Underling: Nothing.
CIO: Then why are we worrying about this again...?
You can see why big companies would rather not redo their entire IPv4 space.
For those who suggest just using NAT and a non-routable/8 - what do you do when you have your first merger with another company that has done the same thing? If Ford merges with Pfizer, they have completely compatible networks already since they are both using their own/8s. If they both use the 10 Class A then they'll have a huge mess.
Frankly, I think the solution is just to go to IPv6 and let anybody who wants to have a class A. Sure, it isn't efficient, but sometimes it is better to have a little virtual waste than to have to plan everything down to the last IP...
Why not have that replacement PC already ghosted and sitting in the closet. Then your client just needs to pull it out and plug it in. I'm sure the owner you do business with can see the value in not needing to pay you to drop whatever you're doing to fix his problem, when all they need to do is pull the cables out of the back of the PC and plug them into the replacement (in a typical PC the cables only fit in one connector - not like much can go wrong). Then, at your leisure (and at non-emergency rates) you can drop by, RMA the bad PC and obtain and ghost a new one. Forget the burnin - Dell already does that for you, but if you're concerned about DOAs you could go ahead and do an overnight burnin before you stick it back in the closet for the next time a PC dies.
This scenario is cheaper, gives a faster turnaround on emergency outages, and lets you handle servincing PCs on a routine basis and not an emergency basis, which is kinder to both your schedule and theirs, as well as to both of your pocketbooks.
I just can't imagine a scenario where you don't care about downtime enough to have fully redundant hardware already running hot, but do care about it enough to pay somebody to run out and swap power supplies in the middle of an outage and build custom-made PCs. If your 5-employee customer loses $1 million an hour for downtime then they should have a spare desk or two with a full PC up and running already so that if something dies they can just run over and resume where they left off. If they lose $20 per hour for a single PC outage then maybe it makes sense to have the spare in the closet.
And yet, somehow, Dell seems to turn a profit on these high failure rate systems.
They have much better deals with their suppliers than what you'll find anywhere at the scale you'd operate. Also, they pay their help a lot less than your customers are paying you.
I build my computers because I like to control the parts that go into them. However, I'm under no illusions that my needs are even remotely similar to a corporate client's needs. They just need a basic reliable system, and this is a commodity market.
You mention high-reliability as one of your goals in building them yourself. How high a reliability to they need? If this is a large business then you just keep a few extra systems in the closet - that has to be a lot cheaper than paying you to put them together (unless you make $8 an hour). The only time it makes sense to pay a premium for reliability is when you're talking servers, and then you can just buy a server hardware line from dell/hp/IBM/whatever. Those have redundancy on most of the parts, and you can get 4-hour 24x7 support contracts, so if something goes it will be fixed that day.
Sometimes it makes sense to do things on your own, but when an entire industry supplies what you're doing, chances are that they're doing it cheaper...
The shuttle retails for $260 The cpu is $80 $50 for ram $100 for the hard drive $70 for the tuner $130 for XP MCE $40 for remote $50 for 9200LE $23 for the sound card Grand total is: $803
If you want mythtv instead you can drop the $130 for XP.
In any case, you're WAY over $400. You also don't have a DVD drive, which makes the DVD playing software a bit unnecessary. So, add another $30-40, unless you want to burn DVDs in which case it is probably higher.
I've debated a mythtv setup, but if I want dual tuners I'm spending almost $1000 with a stack of satellite receivers and a PC. Or, I can get a DVR essentially for free from any cable co. I'd really like to have the freedom that myth offers, but it comes at a hefty price unless you have a PC lying around...
You could save a little with different hardware (no need for separate video card or sound card unless you're playing games). However, you won't save that much. Any PC you build on your own is going to be well over $500...
Often large corporate buyers have a deal with MS that they can install XP Pro or 2K on a workstation as long as it was licensed with any version of windows to start. That often is the best way to leverage the OEM's license with their own buying power. So, Dell puts win98 on the PC, and the megacorp reimages with XP Pro or whatever.
And yes, often the corporate image is loaded by Dell/HP/whatever.
Well, sure a genomicist would have more in common with a biochemist than a chemist, just as a biochemist would have more in common with a chemist than an astronomer. There is essentially a continuum of physical and natural sciences, and biochemistry tends to fall in-between. Biochemistry itself is a pretty broad term, covering everything from splicing genes into plasmids and growing vats of bacteria and churning out recombinant proteins, to studying electrical conductivity of DNA molecules. The former is fairly separated from traditional chemistry, and the latter is almost indistinguisable from pure physical chemistry.
You can find biochemistry programs in biology, medical, and chemistry departments - often in the same university. Each tends to cater to a general region on the physical-biological continuum.
Chemistry is a very broad field, as is biochemistry. And you really can't understand metabolism without a good understanding of kinetics and thermodynamics - two very physical subjects. One way of looking at life is just a bunch of catalysts that carefully control the reaction of oxidizing and reducing agents in their enviornment, using the resulting energy to replicate, get around, and generally do useful things like speculating on what life actually is...:)
Agreed. Engineering can clearly be applied to software - it has parts that interact, and these parts can be characterised and systematically evaluated for suitability for a purpose. However, when you're talking about an OS or a RDBMS or something like that, the number of "parts" enters the 7-8 figures (or even more!). So, engineering on one of these projects is something along the lines of the Apollo program. Sure, you can apply engineering on this scale and get to the moon, but expect to spend a few percent of the US GDP in the process. And Apollo wasn't exactly perfect either (how many people have been killed by an ERP implementation?)...
The fact is that most projects don't need this kind of reliability. As a result we live with scaled-down engineering processes, and obviously this is not adequte to completely remove all bugs.
I couldn't agree more. Unfortunately, the realities of large organizations often are that doing anything quickly is difficult. Obviously the best solution is to update the spec, and have good visibility of the changes across the development community...
As somebody who has worked on fairly large IT projects with very formal specs - that just doesn't happen. Sure, the specs are supposed to be living documents, but in most bureaucratic organizations it is a miracle they get signed off in the first place.
What usually happens is somebody ends up just violating the specs and documenting why they did it, and a note is made to fix the specs whenever it becomes convenient (as in once the project is done and it won't hold anything up).
The problem with the typical project documentation process is that most of the documents never get signed off until the project is almost done, and then the work which actually serves to improve the true quality of the product (which is the software, not the paperwork) gets crammed into the last few months of the timeline.
Don't get me wrong - specs have a place. However, there is a big tendency on the part of managers to emphasize specs (the stuff they can read and understand) over the design/implementation (the stuff only the coders understand), when the latter obviously has more of a direct impact on quality than the former...
I'll read more about this, but both parties are acting unusual to the point where I am really on neither side, whereas normally I suppose I would be on his side.
Uh, according to your explanation if he was acting more normally you'd never have heard of him, let alone take his side. The action you disagree with is his violating a court gag order. If he didn't violate this, nobody would hear about it at all, at least not until the legal system works its self out, he becomes destitute, has all his evidence confiscated, and becomes some old news story that nobody would publish anyway. If he really wants to get the word out, now is probably the time...
But what if the webapp you like was customized from the original GPL code? You can't just run it yourself since the source isn't available.
Yes, you won't have the data, and I don't think anybody is suggesting changing that. However, you will have access to the source in the future, which will be a benefit to everyone. Even if you personally will never access the source to some website running a customized version of a GPL project, the original developers will benefit from being able to merge these customizations back into the original project.
#2 is exactly what this so-called infection is about. The GPL has the nasty habit of making sure that every derived work will be GPL too
Uh, this is in fact the very design of the GPl, not a "nasty habit" (which implies that it is an undesired side-effect). People who write GPL'ed code don't want you to redistribute it in non-GPL projects. If you don't like it take it up with whoever wrote the code - they can in fact grant you an exception. In fact, if you only want to borrow 2 lines of code you just need to figure out who wrote those two lines (not hard with CVS in theory).
If GNU readline would've had a sensible license (like LGPL), there wouldn't have been people working on a replacement library under a less restrictive license. Now they have been, and this time could've been spent better.
It has a prefectly sensible license now. You just happen to not like it. Apparently some other people agree, and they object to the GPL so strongly that they're willing to reinvent the wheel.
Next you'll be arguing that if linux had a sensible license (like BSD) then Microsoft wouldn't have to rewrite their own kernel from scratch for their proprietary OS. I doubt that Linus or RMS could care less - their philosophy would be that MS is free to write free software if they want, or they can compete against it by writing their own software. What neither is interested in is letting MS borrow their code and then enter into direct competition against them.
The GPL is designed to support free (as in freedom) software. It is not intended to boost the public domain (in the legal sense), or to make non-free-as-in-freedom software more popular. Do you think that when RMS wrote emacs his number-1 goal was to make it popular?
Uh, do you really think that RMS cares if a megacorp uses emacs as their primary word processor? That wasn't the reason that he wrote it.
RMS's goal isn't to make FOSS popular in the fastest way possible. RMS believes that non-free software is immoral and wants to ensure that there is moral software out there that he and others of like mind can use. Long term he'd like to stamp non-free software out of existance, by building up such a huge library of free software that nobody would conceive of reinventing the wheel just so they can make it proprietary.
Rmemeber, RMS doesn't just care about people who develop software - he also cares about end-users. The end-user of a website isn't the company that runs the webserver - it is somebody at home with a browser. So, his philosophy would tend to suggest that this user should be able to access the source to the web server.
...it also goes for Peoplesoft and any of the Oracle Applications.
-1 Redundant!:)
Seriously, though, poster is right on. Anybody not familiar with the business IT world should google ERP + lawsuit. These systems are HUGE sinks of money, and rarely do implementations go straightforwardly. The joke with SAP of course is that their salesmen won't talk to anybody whose job title doesn't start with a "C".
No, a scheme which relies on keeping a secret worth billions and known by all to exist is not going to work. The only question is how many creative break-in attempts do we see before one succeeds.
Uh, verisign has had such a scheme for about a decade now. About the closest to a breakin was some guy who got them to issue him a key on behalf of MS, which was quickly revoked. As far as anybody knows, there has been no attempt to obtain the signing key. They could probably defend it from anybody short of an army - I doubt that anybody on their own can even access the computer it physically resides on, and most likely electronic access to the machine is carefully filtered. It is possible to create a completely secure firewall for the connection if it only needs to pass one particular type of data back and forth...
Depends on how it is implemented. If they use one key for all the cards there, it is fairly certain that somebody will crack it and publish it.
On the other hand, if they have one private key kept only by the vendor, the public key for this on each device, a serial number on each device, and a unique private key on each device with a certificate, then it won't be cracked. Sure, DVDJon can crack his flash device, and then he could read/write the data off it. However, your device uses a different key. If he cracks it using software-only, then this could be distributed. More likely, though, he will crack it using logic analyzers and electron microscopes, and you can't exactly just post do-it-yourself instructions for that online. He could mass-produce clones of his card, but the vendor could revoke his key once they found out about it.
I'm not sure how the protection is implemented, but if they really wanted to stop hardware cracking this is exactly how they would do it. Of course, just using one key is easier, and so who knows what they really did...
Thanks for the info. I'd never heard of it. (Probably because I've never used an open-source app that advertised its support for it.) Actually, do any popular open-source apps actually support it? It would seem like a logical standard unless it has some drawback.
I agree that legally I may make my GPL project depend on a BSD-licensed product.
Perhaps suprisingly to you, however, many people would choose not to, given the choice of a GPL competitor. This encourages more widespread use of the GPL. It is the same reason why RMS doesn't use MS Word even though he doesn't have any pressing need to modify its source code.
Some may not care that much about licenses and that is fine. Nobody is putting a gun to anybody's head and forcing them to use mysql when creating a new application.
Now, my one pet-peeve in this area is that there is no linux equivalent of ODBC. If I write a windows app it can be made database-neutral - just use ODBC and it doesn't matter if the datacenter prefers SQL Server, Oracle, or even an FOSS DB. On the other hand, with linux everything is linked to libraries and uses db-specific function calls. Sure, you can modularize your code to make it easier to support multiple DBs, but you're doing all the work of supporting other platforms yourself, and your users are stuck with whatever you're willing to support. Sure SQL itself isn't the most perfect of languages portability-wise with all the vendor-specific extensions, but with ODBC you can generally plug any DB into any app. I really wish that somebody standardized on such a model for linux...
I wonder if GPS signals are authenitcated in any way? If not, the Iraquis could do the same thing (redirecting bombs). Of course, US bombs would use the encrypted military channel, so that would prevent Iraqi spoofing (although the US could spoof those bombs if they desired).
Slashdot Mirror
You also need a system that will allow invalidation of votes, and an inability to prove how you voted after the election.
This way when your boss asks you to vote a certain way on the job you can go ahead and do so, knowing you can change your vote later. And the next day you can't prove you voted one way or another so the boss is none the wiser.
You think big corporations control congress now? Wait until they literally hold tens of thousands of actual votes, and the ability to pay people for their votes (come into this booth with an unused voting card and we'll give you $10!).
Your contrived example is BS. All of these /8's have been allocated for years. The organizations holding them have had a decade to implement best practices and move away from such large, unnecessary utilizations of address space.
/8 network) to avoid investing in complex address assignment schemes.
You make a value statement that it is better to not "waste" space.
If you look at the homes on your street (if you live in the US) they are probably not numbered sequentially. They probably skip every 10 numbers at least. This is so that if somebody puts up a new house you don't need to renumber the entire street. Sure, it is a waste of numbers, but numbers don't cost anything. It is more of a waste to not design for expansion.
In the same way these corporations are being smart in using their address space heirarchically, leaving room for growth. They are using a resource that is very cheap and plentiful (address space on a
These companies would care no more about renumbering their networks than a town might care to renumber their street addresses, even if it were practical to do so. Numbers are free - why not use them?
Staying with IPv4 is like solving the Y2K crisis by going to 99 month years, 99 day months, 99 hour days, 99 minute hours, and 99 second minutes. Sure, it prolongs the agony, but it makes no sense and ultimately doesn't solve the problem. Why not just start rolling out IPv6 where it is easy to do so? Design OS'es so that they obtain IPv6 addresses automatically via DHCP if they are available and use them, design routers so they give them out automatically, and eventually as stuff is replaced everybody will be running on IPv4 and IPv6. Then you can simply stop using IPv4 and nobody will notice. There is no reason to make some grand proclaimation that on 31-DEC-2007 we'll suddenly stop using v4 and start using v6.
Often the best solutions to problems are non-central ones. Sure, it isn't efficient in terms of address-space usage, but when you have more addresses than electrons in the universe should we really be concerned with that? To me it sounds like a bunch of IT guys standing in ivory towers saying that 32-bit addresses ought to be enough for anyone...
Funny you mention it. While not sharing in the analog partyline sense, PBX's already do this very thing. How many extentions are there on your office phone system? 20, 50, 100? How many actual CO lines are there? 8, 20? It's very rare for a PBX to have one CO line for each extention. In fact, that's counter productive -- just put a damned POTS phone on each desk. (Telco's used to sell this as centrex service.)
And likewise a corporate branch office I work at has about 25-30 globally routable addresses, and yet only one line running to and from it. That doesn't change the fact that it has globally-routable addresses. The line goes back to another corporate office, and only accesses the internet at large via proxy servers.
And there's equally no reason why they should. Do you really need to ping the nerf darts on my desk?
No, but perhaps you might have a need to do so when you are't in the office?
One major annoyance of mine is the difficulty in accessing home computers remotely due to NAT. This is a result of the general mindset among ISPs that IP addresses are a resource to be hoarded, and that only broadcasters should really have static ones, and everybody else should just be happy to use their interweb-TV.
The power of the Internet is P2P. However, P2P cannot operate correctly over NAT - at least not if there is NAT at both ends. As soon as you add a central server to proxy connections it is no longer truly P2P.
My whole point is that it doesn't cost a whole lot to allow everybody to just have their IP addresses, so why not just do it? Why should there be some burden of proof that I really need those addresses, as if I should care whether anybody else thinks I really need them or not. It would be one thing if they cost money to manufacture, but we're talking about numbers - they all exist already, we just need to start using them...
Franky, I tend to take the opposite appraoch. I make a point not to check voicemail often, unless I know somebody was planning on calling me.
If somebody has a problem, they can write me an email. I still won't read it right away, but it will get a response. In their email they will have had an opportunity to spell out their problem clearly, so in my response they'll get an actual answer. Typically when people leave a voicemail they just ask you to get back to them so they can explain their problem. I will usually just send them an email asking them to explain their problem.
If there is some kind of system outage it will be apparent from contacts from the official support organization, or by a flurry of phone calls. However, EVERYBODY wants their question answered right away, and I exercise extreme discretion at answering the phone as a result.
My customers have always evalulated me positively, and I think that it is because I meet their real needs. If customer management says that priority 1 is getting a certain project done, they don't want me to drop whatever I'm doing every time a subordinate wants to ask a question about a lower priority project. I allow my customer management to set my priorities, and I conciously apply my efforts where they make the most difference. This way, the largest number of people within the organization benefit.
Simply answering the phone every time it rings is allowing any random person to dictate your schedule. That is not likely to lead to effective use of time.
I already know your internal IPs. They have the form 192.160.x.y, or maybe 10.x.y.z. No reason that somebody can't write a virus that infects one border machine and then scans away at your internal network.
Most major corporations have routable addresses inside their networks - they don't hurt, so why not have them? They put up a nice firewall on all their entrances, and then it is just as secure as NAT (which is still only moderately secure - I'm sure NAT didn't protect many from Sasser and such - it still just takes one person with a laptop).
In most of these companies the internal extensions on the PBX are also world-dialable. It is cheaper than having a horde of operators to redirect calls, and more convenient.
Plus, when you have a corporate merger everybody has compatible phone numbers and IP addresses.
You speak of everything having an IP address as "pie in the sky". The fact is that if ISPs just started assigning routing IPv6 in addition to IPv4 it would just be a matter of time before everything transitioned. There is no reason everything shouldn't have its own IP address.
I'm sure back in the day of party lines the concept of having your own phone number was "pie in the sky". In any office only 10% of the people are on the phone at any time - why not just assign 1 phone number to each 10 desks and ask them to share...
Ok, you're assuming your IP addresses will change in the first place.
/16 or something like that. If you switch ISPs you can just keep the same IPs since you own them.
/16 you just need a router between them and no extra setup.
If you are a company with 50 computers, then you're just getting your IP space from your ISP, and that will be the case. As long as you have DHCP you'll still be fine. And for those two office copiers that don't support it you'll only need 2 minutes to fix them.
Migrations are only a headache when you have thousands of computers. In that case, you probably don't need to use your ISPs address space, but instead you can just get your own
Here is a problem with NAT - mergers. What happens when two companies merge that have private networks? You suddenly have thousands of networks with the same addresses to remap. If on the other hand each company had a
In an ideal world, one machine = 1 IP, globally routable. Sure, we use tricks when this isn't easy to implement (you can't very well have home dialup users taking their IPs with them when they switch ISPs unless you want the top-level routers to have to track every individual IP on the planet). However, NAT is really just a hack and is rarely the best solution to a problem...
Ok, while clearly most of these companies don't need a /8, envision yourself as the CIO of said company.
/16 or two for our /8.
/12. They gave each building a /16, those buildings gave each floor a /24 or two. Plus, we have a bunch of /16s and /12s for various testing purposes and dedicated private networks. We'd have to reprogram every router in the company - we only have about 500 of those. Then we'd have to do all kinds of software testing, and a bunch of systems with static IPs would need new ones.
/8 we had space to spare and could just parse out large blocks to site heads, who could delegate it as they saw fit. With a /16 we need central management of all addresses so that we don't waste too much space. We need some system for keeping track of our addresses now...
/8s. If they both use the 10 Class A then they'll have a huge mess.
Underling: The public just called and wants to trade us a
CIO: Doesn't sound bad on the surface we don't have nearly that many systems. What is the downside.
Underling: Well, we gave every worldwide site their own
CIO: That doesn't sound too hard. Get started on it.
Underling: Uh, how do you want to parse the IPs out - with the
CIO: Uh, I wasn't too good at this stuff in school - I was better at the marketing classes and learning how to go to dinner with vendor sales reps. Go on...
Underling: No problem, we just need a new web-based application that will use a database to...
Secretary: Excuse me, CIO, your golf appointment with the CEO is in 5 minutes.
CIO: Oh, I gotta run. How much money will we make by completely revaming our whole network?
Underling: Nothing.
CIO: Then why are we worrying about this again...?
You can see why big companies would rather not redo their entire IPv4 space.
For those who suggest just using NAT and a non-routable/8 - what do you do when you have your first merger with another company that has done the same thing? If Ford merges with Pfizer, they have completely compatible networks already since they are both using their own
Frankly, I think the solution is just to go to IPv6 and let anybody who wants to have a class A. Sure, it isn't efficient, but sometimes it is better to have a little virtual waste than to have to plan everything down to the last IP...
Why not have that replacement PC already ghosted and sitting in the closet. Then your client just needs to pull it out and plug it in. I'm sure the owner you do business with can see the value in not needing to pay you to drop whatever you're doing to fix his problem, when all they need to do is pull the cables out of the back of the PC and plug them into the replacement (in a typical PC the cables only fit in one connector - not like much can go wrong). Then, at your leisure (and at non-emergency rates) you can drop by, RMA the bad PC and obtain and ghost a new one. Forget the burnin - Dell already does that for you, but if you're concerned about DOAs you could go ahead and do an overnight burnin before you stick it back in the closet for the next time a PC dies.
This scenario is cheaper, gives a faster turnaround on emergency outages, and lets you handle servincing PCs on a routine basis and not an emergency basis, which is kinder to both your schedule and theirs, as well as to both of your pocketbooks.
I just can't imagine a scenario where you don't care about downtime enough to have fully redundant hardware already running hot, but do care about it enough to pay somebody to run out and swap power supplies in the middle of an outage and build custom-made PCs. If your 5-employee customer loses $1 million an hour for downtime then they should have a spare desk or two with a full PC up and running already so that if something dies they can just run over and resume where they left off. If they lose $20 per hour for a single PC outage then maybe it makes sense to have the spare in the closet.
Just food for thought...
And yet, somehow, Dell seems to turn a profit on these high failure rate systems.
They have much better deals with their suppliers than what you'll find anywhere at the scale you'd operate. Also, they pay their help a lot less than your customers are paying you.
I build my computers because I like to control the parts that go into them. However, I'm under no illusions that my needs are even remotely similar to a corporate client's needs. They just need a basic reliable system, and this is a commodity market.
You mention high-reliability as one of your goals in building them yourself. How high a reliability to they need? If this is a large business then you just keep a few extra systems in the closet - that has to be a lot cheaper than paying you to put them together (unless you make $8 an hour). The only time it makes sense to pay a premium for reliability is when you're talking servers, and then you can just buy a server hardware line from dell/hp/IBM/whatever. Those have redundancy on most of the parts, and you can get 4-hour 24x7 support contracts, so if something goes it will be fixed that day.
Sometimes it makes sense to do things on your own, but when an entire industry supplies what you're doing, chances are that they're doing it cheaper...
The shuttle retails for $260
The cpu is $80
$50 for ram
$100 for the hard drive
$70 for the tuner
$130 for XP MCE
$40 for remote
$50 for 9200LE
$23 for the sound card
Grand total is:
$803
If you want mythtv instead you can drop the $130 for XP.
In any case, you're WAY over $400. You also don't have a DVD drive, which makes the DVD playing software a bit unnecessary. So, add another $30-40, unless you want to burn DVDs in which case it is probably higher.
I've debated a mythtv setup, but if I want dual tuners I'm spending almost $1000 with a stack of satellite receivers and a PC. Or, I can get a DVR essentially for free from any cable co. I'd really like to have the freedom that myth offers, but it comes at a hefty price unless you have a PC lying around...
You could save a little with different hardware (no need for separate video card or sound card unless you're playing games). However, you won't save that much. Any PC you build on your own is going to be well over $500...
Often large corporate buyers have a deal with MS that they can install XP Pro or 2K on a workstation as long as it was licensed with any version of windows to start. That often is the best way to leverage the OEM's license with their own buying power. So, Dell puts win98 on the PC, and the megacorp reimages with XP Pro or whatever.
And yes, often the corporate image is loaded by Dell/HP/whatever.
Well, sure a genomicist would have more in common with a biochemist than a chemist, just as a biochemist would have more in common with a chemist than an astronomer. There is essentially a continuum of physical and natural sciences, and biochemistry tends to fall in-between. Biochemistry itself is a pretty broad term, covering everything from splicing genes into plasmids and growing vats of bacteria and churning out recombinant proteins, to studying electrical conductivity of DNA molecules. The former is fairly separated from traditional chemistry, and the latter is almost indistinguisable from pure physical chemistry.
:)
You can find biochemistry programs in biology, medical, and chemistry departments - often in the same university. Each tends to cater to a general region on the physical-biological continuum.
Chemistry is a very broad field, as is biochemistry. And you really can't understand metabolism without a good understanding of kinetics and thermodynamics - two very physical subjects. One way of looking at life is just a bunch of catalysts that carefully control the reaction of oxidizing and reducing agents in their enviornment, using the resulting energy to replicate, get around, and generally do useful things like speculating on what life actually is...
Agreed. Engineering can clearly be applied to software - it has parts that interact, and these parts can be characterised and systematically evaluated for suitability for a purpose. However, when you're talking about an OS or a RDBMS or something like that, the number of "parts" enters the 7-8 figures (or even more!). So, engineering on one of these projects is something along the lines of the Apollo program. Sure, you can apply engineering on this scale and get to the moon, but expect to spend a few percent of the US GDP in the process. And Apollo wasn't exactly perfect either (how many people have been killed by an ERP implementation?)...
The fact is that most projects don't need this kind of reliability. As a result we live with scaled-down engineering processes, and obviously this is not adequte to completely remove all bugs.
Quality, speed, cost - pick any two...
I couldn't agree more. Unfortunately, the realities of large organizations often are that doing anything quickly is difficult. Obviously the best solution is to update the spec, and have good visibility of the changes across the development community...
As somebody who has worked on fairly large IT projects with very formal specs - that just doesn't happen. Sure, the specs are supposed to be living documents, but in most bureaucratic organizations it is a miracle they get signed off in the first place.
What usually happens is somebody ends up just violating the specs and documenting why they did it, and a note is made to fix the specs whenever it becomes convenient (as in once the project is done and it won't hold anything up).
The problem with the typical project documentation process is that most of the documents never get signed off until the project is almost done, and then the work which actually serves to improve the true quality of the product (which is the software, not the paperwork) gets crammed into the last few months of the timeline.
Don't get me wrong - specs have a place. However, there is a big tendency on the part of managers to emphasize specs (the stuff they can read and understand) over the design/implementation (the stuff only the coders understand), when the latter obviously has more of a direct impact on quality than the former...
I'll read more about this, but both parties are acting unusual to the point where I am really on neither side, whereas normally I suppose I would be on his side.
Uh, according to your explanation if he was acting more normally you'd never have heard of him, let alone take his side. The action you disagree with is his violating a court gag order. If he didn't violate this, nobody would hear about it at all, at least not until the legal system works its self out, he becomes destitute, has all his evidence confiscated, and becomes some old news story that nobody would publish anyway. If he really wants to get the word out, now is probably the time...
But what if the webapp you like was customized from the original GPL code? You can't just run it yourself since the source isn't available.
Yes, you won't have the data, and I don't think anybody is suggesting changing that. However, you will have access to the source in the future, which will be a benefit to everyone. Even if you personally will never access the source to some website running a customized version of a GPL project, the original developers will benefit from being able to merge these customizations back into the original project.
#2 is exactly what this so-called infection is about. The GPL has the nasty habit of making sure that every derived work will be GPL too
Uh, this is in fact the very design of the GPl, not a "nasty habit" (which implies that it is an undesired side-effect). People who write GPL'ed code don't want you to redistribute it in non-GPL projects. If you don't like it take it up with whoever wrote the code - they can in fact grant you an exception. In fact, if you only want to borrow 2 lines of code you just need to figure out who wrote those two lines (not hard with CVS in theory).
If GNU readline would've had a sensible license (like LGPL), there wouldn't have been people working on a replacement library under a less restrictive license. Now they have been, and this time could've been spent better.
It has a prefectly sensible license now. You just happen to not like it. Apparently some other people agree, and they object to the GPL so strongly that they're willing to reinvent the wheel.
Next you'll be arguing that if linux had a sensible license (like BSD) then Microsoft wouldn't have to rewrite their own kernel from scratch for their proprietary OS. I doubt that Linus or RMS could care less - their philosophy would be that MS is free to write free software if they want, or they can compete against it by writing their own software. What neither is interested in is letting MS borrow their code and then enter into direct competition against them.
The GPL is designed to support free (as in freedom) software. It is not intended to boost the public domain (in the legal sense), or to make non-free-as-in-freedom software more popular. Do you think that when RMS wrote emacs his number-1 goal was to make it popular?
Uh, do you really think that RMS cares if a megacorp uses emacs as their primary word processor? That wasn't the reason that he wrote it.
RMS's goal isn't to make FOSS popular in the fastest way possible. RMS believes that non-free software is immoral and wants to ensure that there is moral software out there that he and others of like mind can use. Long term he'd like to stamp non-free software out of existance, by building up such a huge library of free software that nobody would conceive of reinventing the wheel just so they can make it proprietary.
Rmemeber, RMS doesn't just care about people who develop software - he also cares about end-users. The end-user of a website isn't the company that runs the webserver - it is somebody at home with a browser. So, his philosophy would tend to suggest that this user should be able to access the source to the web server.
...it also goes for Peoplesoft and any of the Oracle Applications.
:)
-1 Redundant!
Seriously, though, poster is right on. Anybody not familiar with the business IT world should google ERP + lawsuit. These systems are HUGE sinks of money, and rarely do implementations go straightforwardly. The joke with SAP of course is that their salesmen won't talk to anybody whose job title doesn't start with a "C".
No, a scheme which relies on keeping a secret worth billions and known by all to exist is not going to work. The only question is how many creative break-in attempts do we see before one succeeds.
Uh, verisign has had such a scheme for about a decade now. About the closest to a breakin was some guy who got them to issue him a key on behalf of MS, which was quickly revoked. As far as anybody knows, there has been no attempt to obtain the signing key. They could probably defend it from anybody short of an army - I doubt that anybody on their own can even access the computer it physically resides on, and most likely electronic access to the machine is carefully filtered. It is possible to create a completely secure firewall for the connection if it only needs to pass one particular type of data back and forth...
Depends on how it is implemented. If they use one key for all the cards there, it is fairly certain that somebody will crack it and publish it.
On the other hand, if they have one private key kept only by the vendor, the public key for this on each device, a serial number on each device, and a unique private key on each device with a certificate, then it won't be cracked. Sure, DVDJon can crack his flash device, and then he could read/write the data off it. However, your device uses a different key. If he cracks it using software-only, then this could be distributed. More likely, though, he will crack it using logic analyzers and electron microscopes, and you can't exactly just post do-it-yourself instructions for that online. He could mass-produce clones of his card, but the vendor could revoke his key once they found out about it.
I'm not sure how the protection is implemented, but if they really wanted to stop hardware cracking this is exactly how they would do it. Of course, just using one key is easier, and so who knows what they really did...
Thanks for the info. I'd never heard of it. (Probably because I've never used an open-source app that advertised its support for it.) Actually, do any popular open-source apps actually support it? It would seem like a logical standard unless it has some drawback.
I agree that legally I may make my GPL project depend on a BSD-licensed product.
Perhaps suprisingly to you, however, many people would choose not to, given the choice of a GPL competitor. This encourages more widespread use of the GPL. It is the same reason why RMS doesn't use MS Word even though he doesn't have any pressing need to modify its source code.
Some may not care that much about licenses and that is fine. Nobody is putting a gun to anybody's head and forcing them to use mysql when creating a new application.
Now, my one pet-peeve in this area is that there is no linux equivalent of ODBC. If I write a windows app it can be made database-neutral - just use ODBC and it doesn't matter if the datacenter prefers SQL Server, Oracle, or even an FOSS DB. On the other hand, with linux everything is linked to libraries and uses db-specific function calls. Sure, you can modularize your code to make it easier to support multiple DBs, but you're doing all the work of supporting other platforms yourself, and your users are stuck with whatever you're willing to support. Sure SQL itself isn't the most perfect of languages portability-wise with all the vendor-specific extensions, but with ODBC you can generally plug any DB into any app. I really wish that somebody standardized on such a model for linux...
I wonder if GPS signals are authenitcated in any way? If not, the Iraquis could do the same thing (redirecting bombs). Of course, US bombs would use the encrypted military channel, so that would prevent Iraqi spoofing (although the US could spoof those bombs if they desired).