I have to admit I don't understand why Sun is resisting the switch to Linux. I'm not saying they should dump Solaris over night, but a two or three year transition plan would make a lot of sense.
Get over yourself and your OS. Have you been following the threads recently about how IBM is trying (emphasis on "trying") to turn linux into an enterprise-class OS? Hell, from what I can tell most of their patches are even being refused by Linus, because they will slow down things slightly for the low-end, crappy machines Linux was designed for.
The fact is, even perfect success on low-end, consumer-level hardware being used for geek desktops and low-end web/file/print servers is by no means related to even moderate success on large (starting at 4 procs, but mainly 8 or more) enterprise-class machines, and Linux is nowhere near perfect success on low-end machines, at least for serving purposes.
Linux is great for small things, where your hardware is so cheap that tuning doesn't mean much, but for high-end systems, it's going to be a long time. Go check the recent thread about IBM planning on moving to Linux, and look at what needs to be done to it before it's feasible.
But I think my main point here is, get over yourself, wake up and realize that how you live your life is not necessarily even remotely related to how anyone else lives. The fact that you don't understand why Sun doesn't dump Solaris goes to show that (1) you have never worked with Solaris in an enterprise environment, and (2) you have probably never worked with anything in an enterprise environment.
As to why one might use Solaris x86 instead of Linux, I'm planning on building a dual-proc Compaq this month, and I am going to go with Solaris because I don't want to mess with Linux. It's more of a pain in the ass to install, more of a pain in the ass to secure, and more of a pain in the ass to maintain. I have to decide which distribution makes the compromises that best fit my needs, install it, then configure it so that it actually fits the rest of my needs.
Oh wait, that always fails the first time; now I have to pick a different distribution and see if I can customize it the way I need it, and keep trying until one does.
With Solaris, I install it, install the compilers, download the sources I need, and all of that "just type make and make install" actually works on Solaris, without me having to upgrade binutils, glibc, gtk, the kernel, my filesystem, and everything else.
Yes, I know there are Linux wizards out there for whom it "just works", but I don't have the massive time required, or even the inclination, to get to this point, and Solaris x86 will allow me comparable (or better, because it's dual-proc) performance for what I need, and I get to work with up-and-coming enterprise technology, like LDAP authentication for the OS, without having to figure out how to compile openldap and it's nine required pieces of software, plus all of PADL's tools, plus OpenSSL if I want SSL, plus...
Yeah, I feel strongly about this. Linux is not ready for the enterprise; the sooner the Linux community accepts this, the sooner the problem will be fixed. Just like the fact that Linux is not ready for the desktop; as soon as people realize that the majority of users hate spending 20 hours configuring their stupid computers, Linux hackers will actually try to fix it. Until the community acknowledges the problems, though, instead of holding them up like badges, I wouldn't expect to see Sun dumping Solaris.
The fact is, x86 systems are the only ones left that take this stinking long to go through POST (power-on self-test). Yeah, a $300,000 Sun box takes longer, but its POST is _significantly_ more complicated (try it in verbose mode sometime if you don't believe me). Check out how long it takes a Mac to start booting the OS--oh, that's right, it's barely even discernable!
On my machine, a dual-celeron with SCSI, on-board HPT controller, three SCSI devices, one IDE device, and 384MB of RAM, it takes longer to get through the BIOS than it does to boot my main OS, BeOS (of course, Linux is still slower than the BIOS).
I think that's silly. It's not like it can't be done faster, it's just that the x86 world depends on such stinking old technology that it isn't done faster. Everytime I see phrases like "extended RAM" and "640KB" I want to puke. Why haven't we dropped that like the dead weight it is?
So basically, yeah, I want this, and I want it five years ago. I know that this product is for embedded devices, but I want it for my desktop and laptop. Especially laptops--I hate watching the BIOS churn for twenty seconds on a laptop when I am constantly turning it on and off on trips--yucko wasted power.
When are we going to see something like Sun's OpenBoot as our BIOS?
Thats fine for the comand line, but it doesn't work so well for interactive editors.
Actually, it's not fine for the command line, either. Most people seem to forget that your (local) typing does not directly cause characters to appear in your (local) terminal; your (local) typing causes your (remote) connection to echo the characters back to your (local) terminal.
Thus, if you turned on linemode in the command line, you would never see what you typed until you hit return.
Obviously, this could be fixed if we completely redesigned how unix sees the world, but...
Probably the best recommendation I have is to first break absolutely everything you can think of, one at a time. Then, when you go to fix each thing, think about how you'd fix it such that you never, ever have to deal with this problem again.
The second part holds true for any task I ever do as a sysadmin. I hate doing things once, much less twice, so I either design problems away at the beginning or I write a script to automate the solution.
The advice above, breaking everything, does a good job of testing/teaching problem solving skills, and the second piece does a good job of forcing you to think about more than your immediate circumstance, which is very important but often different.
I once had to rebuild an entire NIS+ architecture in a crux situation. Instead of going through and trying to do it by hand, I wrote scripts to automate almost everything. Yeah, it seemed like a waste to my boss sitting behind me, but when everything blew up 90% of the way through it, I was able to get back to 90% done again in about ten minutes.
The reason this is good is because solaris actually is used for more than just servers. Sun does actually make workstations, and sells quite a few of them.
For one thing, there are people like me, who maintain Sun systems all day and who essentially require solaris on the desktop. I use solaris every day as my main GUI, and what window manager to I use? Why, that would afterstep. I more modern window manager probably wouldn't be a bad thing.
Yes, I know I could build it myself, but it's nice to see Sun admit that CDE wasn't cutting it, and that they should just find something else. I like that it will be a standard on Solaris, and I like that it will be included, so I don't have to compile it myself.
V. Networking. The key here is that the Sun box only has room for one network card. So, if you need 2 cards, the answer is pretty simple. Both the dell and
VA box have room for two. Also, if you want something other than base 10/100 ethernet, you aint gettin it with the Sun box.
Well, kind of--the Sun box already comes with two NICs built in, so you don't have to even buy a card.
And you can certainly get other than 10/100 with a Sun box--you can get GBit ethernet, hell, you can even get ATM.
VII. Memory. Sun's memory bug is no longer around, so memory is pretty much even ground here. All of these guys come with 128 Megs on the lowend,
which is probably too little depending on your purpose.
There never was a "memory" problem; it was a cache problem, and it only occurred on the 400Mhz procs that came in the UE machines, which the X1 most certainly is not. And you can use standard PC133 ECC RAM in the X1s, so it's cheap to upgrade them.
And for the record, I would do straight performance tests--nothing else matters in this arena, if these are the only machines you are installing. If you already have a big data center, go with what you have. I personally think Solaris is significantly superior to any other OS I've seen for the data center, especially with Jumpstart, consoles, etc.
(Barracuda? I don't seem them using IBMs!)
Actually, Sun uses IBM drives in a lot of their systems. There doesn't appear to be much conflict between them on a parts level.
But I agree, I wish they added a SCSI option to this box; at least you can drop two drives in it now, and I'm sure you could change them to scsi, with some creative rerouting of cable and such.
Actually the biggest difference between the IIe or IIi procs and the standard II proc is that all multiprocessing capability is removed from the proc. Since it only comes on systems without the capability of adding a second proc, it really isn't a loss at all.
I wouldn't expect a IIe or IIi to be any slower than a standard II, assuming the cache is the same size (which it often isn't).
Just so everyone knows, cutting 260 COs will only affect roughly 4% of Covad's customers, even though it is probably 10% or more of their COs.
So obviously, they weren't making very much money on these COs, and it is unlikely that any given person will be affected. Specifically, if you live in anything like a real city, you will almost definitely not be affected.
Slashdot Mirror
I have to admit I don't understand why Sun is resisting the switch to Linux. I'm not saying they should dump Solaris over night, but a two or three year transition plan would make a lot of sense.
Get over yourself and your OS. Have you been following the threads recently about how IBM is trying (emphasis on "trying") to turn linux into an enterprise-class OS? Hell, from what I can tell most of their patches are even being refused by Linus, because they will slow down things slightly for the low-end, crappy machines Linux was designed for.
The fact is, even perfect success on low-end, consumer-level hardware being used for geek desktops and low-end web/file/print servers is by no means related to even moderate success on large (starting at 4 procs, but mainly 8 or more) enterprise-class machines, and Linux is nowhere near perfect success on low-end machines, at least for serving purposes.
Linux is great for small things, where your hardware is so cheap that tuning doesn't mean much, but for high-end systems, it's going to be a long time. Go check the recent thread about IBM planning on moving to Linux, and look at what needs to be done to it before it's feasible.
But I think my main point here is, get over yourself, wake up and realize that how you live your life is not necessarily even remotely related to how anyone else lives. The fact that you don't understand why Sun doesn't dump Solaris goes to show that (1) you have never worked with Solaris in an enterprise environment, and (2) you have probably never worked with anything in an enterprise environment.
As to why one might use Solaris x86 instead of Linux, I'm planning on building a dual-proc Compaq this month, and I am going to go with Solaris because I don't want to mess with Linux. It's more of a pain in the ass to install, more of a pain in the ass to secure, and more of a pain in the ass to maintain. I have to decide which distribution makes the compromises that best fit my needs, install it, then configure it so that it actually fits the rest of my needs.
Oh wait, that always fails the first time; now I have to pick a different distribution and see if I can customize it the way I need it, and keep trying until one does.
With Solaris, I install it, install the compilers, download the sources I need, and all of that "just type make and make install" actually works on Solaris, without me having to upgrade binutils, glibc, gtk, the kernel, my filesystem, and everything else.
Yes, I know there are Linux wizards out there for whom it "just works", but I don't have the massive time required, or even the inclination, to get to this point, and Solaris x86 will allow me comparable (or better, because it's dual-proc) performance for what I need, and I get to work with up-and-coming enterprise technology, like LDAP authentication for the OS, without having to figure out how to compile openldap and it's nine required pieces of software, plus all of PADL's tools, plus OpenSSL if I want SSL, plus...
Yeah, I feel strongly about this. Linux is not ready for the enterprise; the sooner the Linux community accepts this, the sooner the problem will be fixed. Just like the fact that Linux is not ready for the desktop; as soon as people realize that the majority of users hate spending 20 hours configuring their stupid computers, Linux hackers will actually try to fix it. Until the community acknowledges the problems, though, instead of holding them up like badges, I wouldn't expect to see Sun dumping Solaris.
Yes, I want this.
The fact is, x86 systems are the only ones left that take this stinking long to go through POST (power-on self-test). Yeah, a $300,000 Sun box takes longer, but its POST is _significantly_ more complicated (try it in verbose mode sometime if you don't believe me). Check out how long it takes a Mac to start booting the OS--oh, that's right, it's barely even discernable!
On my machine, a dual-celeron with SCSI, on-board HPT controller, three SCSI devices, one IDE device, and 384MB of RAM, it takes longer to get through the BIOS than it does to boot my main OS, BeOS (of course, Linux is still slower than the BIOS).
I think that's silly. It's not like it can't be done faster, it's just that the x86 world depends on such stinking old technology that it isn't done faster. Everytime I see phrases like "extended RAM" and "640KB" I want to puke. Why haven't we dropped that like the dead weight it is?
So basically, yeah, I want this, and I want it five years ago. I know that this product is for embedded devices, but I want it for my desktop and laptop. Especially laptops--I hate watching the BIOS churn for twenty seconds on a laptop when I am constantly turning it on and off on trips--yucko wasted power.
When are we going to see something like Sun's OpenBoot as our BIOS?
Thats fine for the comand line, but it doesn't work so well for interactive editors.
Actually, it's not fine for the command line, either. Most people seem to forget that your (local) typing does not directly cause characters to appear in your (local) terminal; your (local) typing causes your (remote) connection to echo the characters back to your (local) terminal.
Thus, if you turned on linemode in the command line, you would never see what you typed until you hit return.
Obviously, this could be fixed if we completely redesigned how unix sees the world, but...
Probably the best recommendation I have is to first break absolutely everything you can think of, one at a time. Then, when you go to fix each thing, think about how you'd fix it such that you never, ever have to deal with this problem again. The second part holds true for any task I ever do as a sysadmin. I hate doing things once, much less twice, so I either design problems away at the beginning or I write a script to automate the solution. The advice above, breaking everything, does a good job of testing/teaching problem solving skills, and the second piece does a good job of forcing you to think about more than your immediate circumstance, which is very important but often different. I once had to rebuild an entire NIS+ architecture in a crux situation. Instead of going through and trying to do it by hand, I wrote scripts to automate almost everything. Yeah, it seemed like a waste to my boss sitting behind me, but when everything blew up 90% of the way through it, I was able to get back to 90% done again in about ten minutes.
Somebody mod this guy up...
Here's another mirror, of the download, not the streamer.
For one thing, there are people like me, who maintain Sun systems all day and who essentially require solaris on the desktop. I use solaris every day as my main GUI, and what window manager to I use? Why, that would afterstep. I more modern window manager probably wouldn't be a bad thing.
Yes, I know I could build it myself, but it's nice to see Sun admit that CDE wasn't cutting it, and that they should just find something else. I like that it will be a standard on Solaris, and I like that it will be included, so I don't have to compile it myself.
Well, kind of--the Sun box already comes with two NICs built in, so you don't have to even buy a card.
And you can certainly get other than 10/100 with a Sun box--you can get GBit ethernet, hell, you can even get ATM.
VII. Memory. Sun's memory bug is no longer around, so memory is pretty much even ground here. All of these guys come with 128 Megs on the lowend, which is probably too little depending on your purpose.
There never was a "memory" problem; it was a cache problem, and it only occurred on the 400Mhz procs that came in the UE machines, which the X1 most certainly is not. And you can use standard PC133 ECC RAM in the X1s, so it's cheap to upgrade them.
And for the record, I would do straight performance tests--nothing else matters in this arena, if these are the only machines you are installing. If you already have a big data center, go with what you have. I personally think Solaris is significantly superior to any other OS I've seen for the data center, especially with Jumpstart, consoles, etc.
(Barracuda? I don't seem them using IBMs!)
Actually, Sun uses IBM drives in a lot of their systems. There doesn't appear to be much conflict between them on a parts level.
But I agree, I wish they added a SCSI option to this box; at least you can drop two drives in it now, and I'm sure you could change them to scsi, with some creative rerouting of cable and such.
I wouldn't expect a IIe or IIi to be any slower than a standard II, assuming the cache is the same size (which it often isn't).
Yes, it's got sound; if you look at the specs, it's got four audio ports, line in, line out, microphone in, and speaker out.
Should work well, since the audio on the ultra 5 always did.
Just so everyone knows, cutting 260 COs will only affect roughly 4% of Covad's customers, even though it is probably 10% or more of their COs.
So obviously, they weren't making very much money on these COs, and it is unlikely that any given person will be affected. Specifically, if you live in anything like a real city, you will almost definitely not be affected.
Actually, "Covad" stands for "Cable over Voice and Data"; kind of a pun, really, because it's really voice and data over cable.
So no, I don't think they paid someone to think it up, and I don't think it got taken from a government acronym.