Like I said in a previous post, I don't think my experience is atypical. I cannot believe that my shop is alone in not having constant software issues with MS software. Again, it comes down to competency...I'm reminded of someone's sig. here on/. it goes something like clueless admin Homer Simpson on NT: Server goes up, server goes down, server goes up, server goes down....you get the picture.
In any case, Microsoft has *always* given very favorable discounts to Academia, not just when it suits them. But they're not stupid either...giving large discounts to schools and students helps promote another generation weaned on MS software, another generation using/administering/and coding for their software simply because they're used to it. Then MS hopes and prays like hell that when they can pay for a copy, they will.
Also, you're right, I don't think that linux pricing is more expensive. In all of my previous posts I explicitly stated that I couldn't argue that point because it *is* nearly impossible to argue (that said, now that I look at it, the support you get for the $1499 RH AS license looks pretty paltry...9-5 only and only for one machine....)
Anyway, it's late and I'm going to bed. Perhaps I'll continue this another day.
No that's $87 per licensed copy of Windows. So to acquire 100 licenses, yes, that would be $8700. But the support is per incident....so if you had the same issue with all 100 servers, you only need the support once, right? Then you just apply that to the rest of the servers and MS only charges you $250 for the one incident. Now, yes, if all 100 servers had different problems, you would definitely be up the creek without some kind of support contract. Again, I've only had to call MS support once, and they didn't charge me since it was related to a security fix.
...you're figures are off quite a bit for those of us working as admins in the academic world. I realize that you wouldn't know that, and I also realize that most admins do not have access to the kind of pricing that academic institutions can get. I only pay $87 per license of 2K server that I use in a production environment. For academic teaching, we only pay $800 per year for almost everything MS makes, with the exclusion of any Office product. And, I can give a license for all that stuff away to each and every student enrolled in a CS class. We also get 4 free support incidents a year with that annual cost. Now, again, I know that this can't compare with Free Software in terms of absolute price, but I just wanted to make it known that MS's pricing isn't *always* horrible. As a matter of fact, for many universities, their software is downright cheap.
If only 8 things go wrong per year in a 6 server MS shop in both server OS and server apps, you're having a very good year. To expect to have 6 very good years in a row is not very probable.
I'm not going to argue the price issue, but I will say this. I currently run 19 servers: 5 NT 4 and 14 Win2K. About half of the Win2k used to run NT 4 and were upgraded to 2k, the other 7 were fresh installed. I've had 12 of them up and running since 1998, the rest came online in early 2002. In all that time, I have NEVER had to pay for an MS support incident. I did have to call them once for an issue with Exchange 5.5 SP4, but they didn't charge me since it was an issue with a security fix (which, btw, they never charge for support problems with security fixes).
Now, I don't consider myself that lucky, so for someone running 6 servers to go a year, they shouldn't have a problem. Hell, most of my NT 4 servers had >1 year uptimes. So, I don't think it is uncommon for our shop (or any other) to go 4.5 years without a big issue.
knocks on wood, and crosses fingers for next year or so:)
I think the key is directly related to the competency of the sysadmin. Yeah, sure, any kid straight out of HS could probably get a Windows server up and running, but keeping them running and running well does take at least *some* skill and planning, just like it does in the Unix world. It isn't as simple as running the autoupdates or visiting Windows Update every week, but it isn't rocket science either.
The new stand alone Active Directory (application mode AD, as it is called) for apps that require directory service but don't really require a full blown domain. That change alone is worth a major rev. level.
There is also the "restore from media" option that lets you build *new* DCs from the system state backup of an old DC. Previously, you couldn't do that, and bringing up a new DC meant running dcpromo and replicating all the data from the various domains. Big deal you say? An HP IT department had to sync a new DC that was also a global catalog over a WAN line. It took 3 DAYS just for the replication. Obviously this will save some serious amounts of time.
Actually, I'm betting that it would. The difference here is that your friend wanted to go from a superior signal to something inferior...ie the vhs which obviously can't do hdtv res. I'm betting that the output of these things is inferior to a decent traditional speaker setup. Therefore, if the output of the panel was recorded uncompressed with high quality recording equipment, you would be able to discern the quality when played on a normal setup. If, however, the panel output is superior to what you've recorded/replayed it on, then yeah, you're right....it doesn't work.
Yeah, I got my first computer in '89 too...a Tandy Color Computer 2 (lowers head in shame). I didn't get my first real PC until I got a 386/25 in ~1993.
Of course, it seems more cost-effective to just blanket the area with Wi-Fi...
Blanketing the area with Wi-Fi misses the point behind this....ie:easily upgradeable last mile delivery. Current Wi-Fi speeds are great for small areas, but shared 11Mbit (or 54 or whatever) will only last so long. Fiber, however, has nearly unlimited capacity, for all intents and purposes.
Put Wi-Fi in, and you'll be replacing it in 5 or 6 years due to larger bandwidth needs. Use fiber, and in 5-6 years you'll STILL be thinking how to saturate that link.
Oh, and BTW, Cat 5 hasn't been around since '86:)
Just picking! I know what you meant!
I was actually thinking that the clause, as written, seems to outlaw firewalls in general. Not just NAT'ing firewalls, but any firewall, since the point of the typical firewall is to conceal the existence of internal machines.
Not to be an ass or anything, but XP doesn't run on alphas. If you take a careful look at the patch file, you'll see ia64 in the file name. 2 completely different architectures.
And then, there is the HP 1000 at work, about 2 racks in size... is that the ancestor of your box?
I'm not sure, but it sounds logical. The 2 boxes we have now aren't any bigger than a standard tower pc. The 3000 model that we had before (purchased around 1989, can't remember model off-hand) was the size of a dishwasher, and that was just the actual cpu, bus, and other main logic boards. The disk drives were housed externally in another box about the same size as the cpu, and the tape drive was also external, and again about the same size. I'm told that the ancestor of that machine was even larger!
Indeed. Just for fun, we have a test machine that sits outside our firewalls. It is HIGHLY amusing to look at the logs. Of course you can filter out the automated attacks, script kiddies, etc....but occasionally you get the cracker types attempting to figure out what the hell they've stumbled on. We run very weak passwords on it (this box is mostly just for fun, and we could care less if it did get rooted), and still most people can't figure out how to get a remote prompt since we use only VT-MGR protocol and not telnet. Makes for a few laughs anyway.
We still run a couple of HP3000s...a 928 and a 918. Before that we had another model, I forget which one. In any case, there is 20+ years of custom in-house COBOL programming invested in those systems. Most of that code is still serving its purpose very well. We have started updating the apps and have done some web development with it, but if it works, why change it?
The only reason we have even considered migration is because HP has finally pulled the plug on the 3000 line, not because it couldn't serve its purpose. And hey, who doesn't like a half-obscure OS (MPE/ix) running on a 48MHz machine supporting 200+ users?
Win2K w/ Visual C++ 6.0 on a 300Mhz is unbearably slow
Well, yeah:) I didn't say I did anything *serious* on it, other than standard office type productivity. Seriously though, how much memory are you running? Also, try bumping up the compiler process priority to above normal. Should help a bit.
I do run Win2K and Office2k on an older machine...a P200 MMX with 128 MB ram and 2.1 GB disk. It runs fine. Take out even 16 meg of memory though, and forget it. I would try to run it with 256 meg of memory, but the board is so old it only supports 4 32MB simms. YMMV, though.
I'm not a perl programmer, but surely it must have an else statement and your code could be written like so:
sub do_something
{
my $rtc = 1;
if( $value != $correct_value )
{
# clean up stuff
}
else $rtc = 0;
return $rtc;
}
..and it is because MS preloads a lot of the stuff (ie the office startup thing). Even if the office startup isn't loaded, you'll notice that Office apps start much quicker after you've opened them once, shut them down, and then opened them again. It believe it happens because MS simply kills the window, but keeps much of the code in memory so that future start-ups happen much more quickly. Conversely, this can adversely affect other apps that need to start up, since more mem is hogged by the office apps that aren't completely shut down.
When the law was passed, they probably were thinking about ratio servers. However, the wording is sufficiently general enough to be applied to something like Kazaa's user ratings. High download to upload ratio will lower your user rating, and if it gets low enough, you can't download (at least not from the good nodes anyway). So, how to raise the rating (without client hacks, which are available)? Share and upload files. Sounds a lot like the wording of the law to me.
Actually, I saw a blurb on fark.com the other day mentioning that someone set a Raiders vs. Bucs game in Gameday 2003 on a PS2, and (I'm assuming both teams were computer played) the computer predicted a 28-8 Bucs victory.
The scary part is that in the previous 7 times they've done this, they were right every time! Which thoroughly sucks for us Raiders fans! Go Raiders!
If I remember right, the dx included a coproc, but it was off-board, ie: a separate 387 chip. Admittedly, I am a little sketchy on the details right now though, and don't feel like googling for proper support.
Slashdot Mirror
Like I said in a previous post, I don't think my experience is atypical. I cannot believe that my shop is alone in not having constant software issues with MS software. Again, it comes down to competency...I'm reminded of someone's sig. here on /. it goes something like clueless admin Homer Simpson on NT: Server goes up, server goes down, server goes up, server goes down....you get the picture.
In any case, Microsoft has *always* given very favorable discounts to Academia, not just when it suits them. But they're not stupid either...giving large discounts to schools and students helps promote another generation weaned on MS software, another generation using/administering/and coding for their software simply because they're used to it. Then MS hopes and prays like hell that when they can pay for a copy, they will.
Also, you're right, I don't think that linux pricing is more expensive. In all of my previous posts I explicitly stated that I couldn't argue that point because it *is* nearly impossible to argue (that said, now that I look at it, the support you get for the $1499 RH AS license looks pretty paltry...9-5 only and only for one machine....)
Anyway, it's late and I'm going to bed. Perhaps I'll continue this another day.
No that's $87 per licensed copy of Windows. So to acquire 100 licenses, yes, that would be $8700. But the support is per incident....so if you had the same issue with all 100 servers, you only need the support once, right? Then you just apply that to the rest of the servers and MS only charges you $250 for the one incident. Now, yes, if all 100 servers had different problems, you would definitely be up the creek without some kind of support contract. Again, I've only had to call MS support once, and they didn't charge me since it was related to a security fix.
...you're figures are off quite a bit for those of us working as admins in the academic world. I realize that you wouldn't know that, and I also realize that most admins do not have access to the kind of pricing that academic institutions can get. I only pay $87 per license of 2K server that I use in a production environment. For academic teaching, we only pay $800 per year for almost everything MS makes, with the exclusion of any Office product. And, I can give a license for all that stuff away to each and every student enrolled in a CS class. We also get 4 free support incidents a year with that annual cost. Now, again, I know that this can't compare with Free Software in terms of absolute price, but I just wanted to make it known that MS's pricing isn't *always* horrible. As a matter of fact, for many universities, their software is downright cheap.
If only 8 things go wrong per year in a 6 server MS shop in both server OS and server apps, you're having a very good year. To expect to have 6 very good years in a row is not very probable.
:)
I'm not going to argue the price issue, but I will say this. I currently run 19 servers: 5 NT 4 and 14 Win2K. About half of the Win2k used to run NT 4 and were upgraded to 2k, the other 7 were fresh installed. I've had 12 of them up and running since 1998, the rest came online in early 2002. In all that time, I have NEVER had to pay for an MS support incident. I did have to call them once for an issue with Exchange 5.5 SP4, but they didn't charge me since it was an issue with a security fix (which, btw, they never charge for support problems with security fixes).
Now, I don't consider myself that lucky, so for someone running 6 servers to go a year, they shouldn't have a problem. Hell, most of my NT 4 servers had >1 year uptimes. So, I don't think it is uncommon for our shop (or any other) to go 4.5 years without a big issue. knocks on wood, and crosses fingers for next year or so
I think the key is directly related to the competency of the sysadmin. Yeah, sure, any kid straight out of HS could probably get a Windows server up and running, but keeping them running and running well does take at least *some* skill and planning, just like it does in the Unix world. It isn't as simple as running the autoupdates or visiting Windows Update every week, but it isn't rocket science either.
The new stand alone Active Directory (application mode AD, as it is called) for apps that require directory service but don't really require a full blown domain. That change alone is worth a major rev. level.
There is also the "restore from media" option that lets you build *new* DCs from the system state backup of an old DC. Previously, you couldn't do that, and bringing up a new DC meant running dcpromo and replicating all the data from the various domains. Big deal you say? An HP IT department had to sync a new DC that was also a global catalog over a WAN line. It took 3 DAYS just for the replication. Obviously this will save some serious amounts of time.
....nope...not gonna work.
Actually, I'm betting that it would. The difference here is that your friend wanted to go from a superior signal to something inferior...ie the vhs which obviously can't do hdtv res. I'm betting that the output of these things is inferior to a decent traditional speaker setup. Therefore, if the output of the panel was recorded uncompressed with high quality recording equipment, you would be able to discern the quality when played on a normal setup. If, however, the panel output is superior to what you've recorded/replayed it on, then yeah, you're right....it doesn't work.
I didn't even get a CPU at home until 1989!
Yeah, I got my first computer in '89 too...a Tandy Color Computer 2 (lowers head in shame). I didn't get my first real PC until I got a 386/25 in ~1993.
Of course, it seems more cost-effective to just blanket the area with Wi-Fi...
:)
Just picking! I know what you meant!
Blanketing the area with Wi-Fi misses the point behind this....ie:easily upgradeable last mile delivery. Current Wi-Fi speeds are great for small areas, but shared 11Mbit (or 54 or whatever) will only last so long. Fiber, however, has nearly unlimited capacity, for all intents and purposes.
Put Wi-Fi in, and you'll be replacing it in 5 or 6 years due to larger bandwidth needs. Use fiber, and in 5-6 years you'll STILL be thinking how to saturate that link. Oh, and BTW, Cat 5 hasn't been around since '86
I was actually thinking that the clause, as written, seems to outlaw firewalls in general. Not just NAT'ing firewalls, but any firewall, since the point of the typical firewall is to conceal the existence of internal machines.
Not to be an ass or anything, but XP doesn't run on alphas. If you take a careful look at the patch file, you'll see ia64 in the file name. 2 completely different architectures.
And then, there is the HP 1000 at work, about 2 racks in size... is that the ancestor of your box?
I'm not sure, but it sounds logical. The 2 boxes we have now aren't any bigger than a standard tower pc. The 3000 model that we had before (purchased around 1989, can't remember model off-hand) was the size of a dishwasher, and that was just the actual cpu, bus, and other main logic boards. The disk drives were housed externally in another box about the same size as the cpu, and the tape drive was also external, and again about the same size. I'm told that the ancestor of that machine was even larger!
Indeed. Just for fun, we have a test machine that sits outside our firewalls. It is HIGHLY amusing to look at the logs. Of course you can filter out the automated attacks, script kiddies, etc....but occasionally you get the cracker types attempting to figure out what the hell they've stumbled on. We run very weak passwords on it (this box is mostly just for fun, and we could care less if it did get rooted), and still most people can't figure out how to get a remote prompt since we use only VT-MGR protocol and not telnet. Makes for a few laughs anyway.
We still run a couple of HP3000s...a 928 and a 918. Before that we had another model, I forget which one. In any case, there is 20+ years of custom in-house COBOL programming invested in those systems. Most of that code is still serving its purpose very well. We have started updating the apps and have done some web development with it, but if it works, why change it? The only reason we have even considered migration is because HP has finally pulled the plug on the 3000 line, not because it couldn't serve its purpose. And hey, who doesn't like a half-obscure OS (MPE/ix) running on a 48MHz machine supporting 200+ users?
I was wondering when someone would point that out to this guy. Where's the mod points when you need them?
Win2K w/ Visual C++ 6.0 on a 300Mhz is unbearably slow
:) I didn't say I did anything *serious* on it, other than standard office type productivity. Seriously though, how much memory are you running? Also, try bumping up the compiler process priority to above normal. Should help a bit.
Well, yeah
I do run Win2K and Office2k on an older machine...a P200 MMX with 128 MB ram and 2.1 GB disk. It runs fine. Take out even 16 meg of memory though, and forget it. I would try to run it with 256 meg of memory, but the board is so old it only supports 4 32MB simms. YMMV, though.
I'm not a perl programmer, but surely it must have an else statement and your code could be written like so: sub do_something { my $rtc = 1; if( $value != $correct_value ) { # clean up stuff } else $rtc = 0; return $rtc; }
..and it is because MS preloads a lot of the stuff (ie the office startup thing). Even if the office startup isn't loaded, you'll notice that Office apps start much quicker after you've opened them once, shut them down, and then opened them again. It believe it happens because MS simply kills the window, but keeps much of the code in memory so that future start-ups happen much more quickly. Conversely, this can adversely affect other apps that need to start up, since more mem is hogged by the office apps that aren't completely shut down.
...just share everything on a P2P network. Then, after a crash, just fire up your favorite client and go get your invaluable porn^H^H^H^H data files!
When the law was passed, they probably were thinking about ratio servers. However, the wording is sufficiently general enough to be applied to something like Kazaa's user ratings. High download to upload ratio will lower your user rating, and if it gets low enough, you can't download (at least not from the good nodes anyway). So, how to raise the rating (without client hacks, which are available)? Share and upload files. Sounds a lot like the wording of the law to me.
End-of-life? What's that? I mean, c'mon I've still got some apps that run on DOS 3.3 for christ's sake! And that's no joke either (unfortunately).
Ding Ding! We have a winner! Such a funny movie, that was.
Boy, I wonder how many people will get that reference. Hehe, hats off to you though for coming up with it...wish I had!
Actually, I saw a blurb on fark.com the other day mentioning that someone set a Raiders vs. Bucs game in Gameday 2003 on a PS2, and (I'm assuming both teams were computer played) the computer predicted a 28-8 Bucs victory.
The scary part is that in the previous 7 times they've done this, they were right every time! Which thoroughly sucks for us Raiders fans! Go Raiders!
If I remember right, the dx included a coproc, but it was off-board, ie: a separate 387 chip. Admittedly, I am a little sketchy on the details right now though, and don't feel like googling for proper support.