If this is true, and Cygnus is a financially healthy company, then why on earth are they being bought out by the smaller, less profitable(as of right now), RedHat?
Remember, the financial world is mysterious to those of us impaired by a strong attachment to our logical faculties.
For example, when Worldcom bought MCI it was a $8M company buying a $20M company. Somehow these things work.
Having just designed a large server farm, I can testify to the usefulness of Serial as a fall-back remote access channel.
Amen! I use Linux without monitor/keyboard/mouse in phone co pops, rackmounted. I have a console server attach to the serial port, and that's how I get console. How the heck am I going to do that with USB? I sure hope the answer they give is better than "Well, open up the 200 machines you need to deploy this wek and put this handy $300 console board in".
I have also seen some manufacturers using a proprietary protocol to get console, but you need to have an NT machine on the local subnet for it to work. (as a side note, they say the protocol runs over ethernet and talks to your box even if it's powered-down, which just creeps me out)
I hope the x86 manufacturers are contemplating these kinds of questions, otherwise they risk giving this part of the business up to Sun etc. forever.
This is just a lone opinion, but remember back when the MCI-Worldcom deal was on the table? Regulators were jittery from the outset because they knew that without a proper divestiture of MCI's internet assets before the merger close, the combined company would end up controlling more than 70% of the backbone business.
I think about this latest proposal the same way. You have one company in control of two major backbones (UUNET and Sprint), not to mention their own service, ONNet. I would venture to guess that this combined company would control a minimum 60% of the backbone biz, without any divestiture of Sprint's internet business. That just isn't gonna happen, folks. I can't see another large backbone being spun off like MCI's infrastructure was to Cable and Wireless. I know a lot of you will probably see this differently, but that's my take...
I worked at InternetMCI during the merger, and so I may have some light to shed on the subject.
The regulators (US & Euro) define the Internet business this way: There are "Tier-1" providers, three or four of them. "Tier-1" is defined as follows: Any traffic between any given two points on the internet must contact or traverse a tier-1 provider at some intermediary point. It's laughably absurd, I know, but that's how they think of it. And based on this they identified three Tier-1 providers three or so years ago: UUnet, MCI, and Sprint.
I do not think that this merger will happen without a divestiture of the Sprint backbone for the same reasons that the MCI spinoff to Cable & Wireless happened. More chaos will insue, although this time they will likely spin it off to an American company I would guess. C+W pays my paychecks, but I've always thought the choice of them was just a sop to the European regulators.
One interesting note: The European regulators were more concerned about monopoly in nthe Internet business than the Americans. Not monopoly in Europe, where Uunet had some holdings but MCI was largely absent. But because of those holdings Uunet had, the EC has the right to review the entire merger and block it if they find it will create monopolies that will exist entirely within American territory.
A lot of people have been expressing their suspicion of Sun's motives with regard to the SCSL. But I think that what SUN is trying to do is to codify in a legally defensible position what the BSDs have codified in custom.
One of the virtues of, for example, FreeBSD is not just that it had many contributors, but that all code is checked by a core team of members before it is integrated, and that same core team controls the general direction of the OS (what threads implementation to use, ELF vs. a.out, etc.). Worries about code forking and other problems are a subset of this concern over control.
So yes, they're a big bad corporation, and maybe they're not doing this just because it's a Good Thing. Their worst-case scenario is for their OS, the only OS to run on their high end hardware, to integrate sizable contributions from the open source community that they later discover are buggy crap. So they need to interpose themselves as a middle layer (core team) to guarantee quality, and they can;t rely on just custom to do it, they need a legally binding agreement.
This confirms what we were able to decypher from the patent document. They are building a sort of universal processor capable of translating instructions of other machines into an internal native format and then executing them.
I suppose the interesting thing would be to run benchmarks and see what emplated instruction set runs fastest. A benchmark between Linux/Alpha,/Sparc, and/Intel would be interesting.
I suppose it's just a matter of finding which instruction set is most easily optimized.
Slashdot Mirror
Remember, the financial world is mysterious to those of us impaired by a strong attachment to our logical faculties.
For example, when Worldcom bought MCI it was a $8M company buying a $20M company. Somehow these things work.
Amen! I use Linux without monitor/keyboard/mouse in phone co pops, rackmounted. I have a console server attach to the serial port, and that's how I get console. How the heck am I going to do that with USB? I sure hope the answer they give is better than "Well, open up the 200 machines you need to deploy this wek and put this handy $300 console board in".
I have also seen some manufacturers using a proprietary protocol to get console, but you need to have an NT machine on the local subnet for it to work. (as a side note, they say the protocol runs over ethernet and talks to your box even if it's powered-down, which just creeps me out)
I hope the x86 manufacturers are contemplating these kinds of questions, otherwise they risk giving this part of the business up to Sun etc. forever.
I think about this latest proposal the same way. You have one company in control of two major backbones (UUNET and Sprint), not to mention their own service, ONNet. I would venture to guess that this combined company would control a minimum 60% of the backbone biz, without any divestiture of Sprint's internet business. That just isn't gonna happen, folks. I can't see another large backbone being spun off like MCI's infrastructure was to Cable and Wireless. I know a lot of you will probably see this differently, but that's my take...
I worked at InternetMCI during the merger, and so I may have some light to shed on the subject.
The regulators (US & Euro) define the Internet business this way: There are "Tier-1" providers, three or four of them. "Tier-1" is defined as follows: Any traffic between any given two points on the internet must contact or traverse a tier-1 provider at some intermediary point. It's laughably absurd, I know, but that's how they think of it. And based on this they identified three Tier-1 providers three or so years ago: UUnet, MCI, and Sprint.
I do not think that this merger will happen without a divestiture of the Sprint backbone for the same reasons that the MCI spinoff to Cable & Wireless happened. More chaos will insue, although this time they will likely spin it off to an American company I would guess. C+W pays my paychecks, but I've always thought the choice of them was just a sop to the European regulators.
One interesting note: The European regulators were more concerned about monopoly in nthe Internet business than the Americans. Not monopoly in Europe, where Uunet had some holdings but MCI was largely absent. But because of those holdings Uunet had, the EC has the right to review the entire merger and block it if they find it will create monopolies that will exist entirely within American territory.
One of the virtues of, for example, FreeBSD is not just that it had many contributors, but that all code is checked by a core team of members before it is integrated, and that same core team controls the general direction of the OS (what threads implementation to use, ELF vs. a.out, etc.). Worries about code forking and other problems are a subset of this concern over control.
So yes, they're a big bad corporation, and maybe they're not doing this just because it's a Good Thing. Their worst-case scenario is for their OS, the only OS to run on their high end hardware, to integrate sizable contributions from the open source community that they later discover are buggy crap. So they need to interpose themselves as a middle layer (core team) to guarantee quality, and they can;t rely on just custom to do it, they need a legally binding agreement.
I wouldn't have done it any other way.
I suppose the interesting thing would be to run benchmarks and see what emplated instruction set runs fastest. A benchmark between Linux/Alpha, /Sparc, and /Intel would be interesting.
I suppose it's just a matter of finding which instruction set is most easily optimized.
Richard Simmons?