Standard 15k currently support 72 CPUs normally (4 CPUs in each of 18 system boards). This can be expanded to 106 by putting CPUs in I/O slots, although those CPUs are partially crippled by latencies/bandwith constraints by doing so. USIV give a form-factor equivalent to USIII so it should allow a simple doubling of capacity, assuming the firmware and OS support it.
As for rack sizes, the 15K racks are about the same size as normal racks, but are slightly deeper. The system is not like a standard rack + servers, you buy a 15K as a single unit and add the system/IO boards.
Most people when they think of nuclear power worry about leaks, failures and "3 eyed fish" etc. However, those problems are fairly rare, but the issues regarding long-term nuclear waste storage are, in my opinion, more important. We're dealing with stuff that will be dangerous 100,000 years down the line.
Yup, my first thoughts on depleted uranium was that it was "depleted", therefore not radioactive. Once I read a bit more, I discovered the other issues, i.e. DU is poisonous to the human system if ingested/breathed in and it's a bit difficult no to breathe in dust which surrounds an area after an attack. Basically, on impact, DU superheats (part of its effectiveness) and ends up as a dust residue in the area. According to some sources, this has been causing health problems in both Iraq (after the first war) and the Balkans.
In short, holding a piece of DU is fairly safe; breathing in DU dust isn't.
Re:The customer's needs can be accomplished many w
on
Does IT Matter?
·
· Score: 1
Exactly, IT spending (well, all spending really) should come off a fairly simple cost/benefit analysis. If a project will generate more revenue or garner a reduction in costs greater than the project cost, it should be implemented. Some things are a necessity (e.g. for legal reasons or Health & Safety) but in essence, this comes down to the cost/benefit; the benefit is that you meet legal requirements and don't kill your staff:)
Having worked in a large corporation, budgets are one of the biggest roadblocks to this kind of thinking for two reasons: firstly, it's nigh on impossible to get extra funding even when it would give a benefit if you're out of budget and secondly, budgets lead to an end of year splurge to use up any remaining funds. "Accountants know the cost of everything and the value of nothing" sounds silly until you've worked in that kind of enviroment.
The point of bayesian filtering is that it gets trained by the users; it's most effective at a granular level (i.e. per user) rather than aggregated (i.e. across thousands of users).
The main problem is that 90% of people don't know about filtering; they use whatever software their ISP provides. If they downloaded Mozilla or one of the POP3 filters, they'd get less spam but they just haven't heard of them.
Last I heard, Oracle still has raw disk as an option, mainly for performance reasons but also for parallel Oracle. With direct I/O or Quick I/O (Veritas), you can get near the performance of raw devices, but raw disk is still marginally faster.
Depends what you're serving; a small web site can be 5 MB, NFS/SMB sharing only benefits from memory if you get a lot of reads on the same files so filesystem cache comes in.
Yes, more RAM is nice, but provided your OS doesn't take up most of it, you can do a fair bit in 128MB.
The problem is nothing to do with SIDs. The problem is that there is no way to back up a live (i.e. running) Windows XP (or NT, or 2000) system on NTFS in such a way that you can restore it to a fully functional system without going through (a) installation of XP/2k/NT (b) installation of all software packages individually and then (c) going through the hassle of applying the multitude of security patches on Windows and every package. This probably entails about 10+reboots and a considerable period of time. NTFS locks too many files which are required by a running system (pagefile.sys, registry, etc, etc, etc).
Backing up linux/Unix is a matter of backing up the filesystems with tar/dump/ufsdump/whatever to tape. Restoration involves booting from CD, reparitioning the disk and restoring from tape, installing a boot block and rebooting. If you're being picky, you shut down apps first and backup from single user mode, but 99% of companies get by fine with live backups (other than Oracle and DBs etc).
As far as I'm aware, the only way you can possibly back up a live NT system fully is to run disk mirroring and break off one mirror and back up that using some drive imaging. Restoration involves a complete restore of the entire filesystem, however.
The US invasion of Iraq was a godsend of Al Qaeda in Iraq. It allowed them an environment where they could point at US/Western oppression of the Arabs and allow them to recruit. The chaos in Iraq has also enabled its operatives to slip in and cause havoc against allied troops.
The allies pretty much admitted that Al Qaeda wasn't in Iraq before the war, but they didn't make much of a noise about that because it served their purpose to have the public believe that Al Qaeda was in Iraq to bolster support for the war.
As for N Korea, Bush claiming it was part of his "Axis of Evil" didn't help. NK has now seen what has happened to one third of that (Iraq) and is now trying to make sure it isn't the next target. At this point, you can't really blame it for developing a nuclear deterrant.
*sigh* Clusters come nowhere near the level of fault tolerance you get in big iron. In a real fault-tolerant system, there are multiple paths for all transactions; in essence, you're running the same code on two (or more) CPUs. If one fails, you have zero downtime, other than a quick reroute in hardware to compensate (you probably wouldn't even notice it). To the best of my knowledge, there is no clustering solution which comes close to this, whether based on linux, Unix or Windows.
Yes, clusters can do a job which is "good enough" to replace expensive mainframes, but there are some cases where they aren't good enough, especially banking where you have to be 100% confident that every transaction is logged correctly.
Many people haven't switched yet, for a number of reasons. Personally, I'm used to 1.3.x so I'm sticking with it, but the big killer for most sites is the lack of mod_perl and php other than betas.
FWIW, the server name is transmitted in a standard HTTP/1.1 response so it's trivial to work out what kind of server something is running. As a simple test, run 'telnet [host] 80' and type 'GET / HTTP/1.1' and hit enter a few times. You'll get a response (usually an error saying invalid HTTP/1.1 request) which includes a server version.
Yup, it would be difficult. Especially at high altitude, as the plane undergoes severe decompression. There's a very good reason not to allow guns on aircraft, as firing a gun in an aircraft cabin will likely lead to an imminent crash.
This has to be the best advantage of subscription; you get to beat the rush to view the site before the stampeding hordes of/.ers kill the poor server:)
Ah, if you use multiplexing then yes; OTOH, you can have multiple tape drives and have Netbackup stream one client to each tape drive and that will also speed up restores (it will be able to stream from tape rather than stream/seek). This is true for all backup products, not just Veritas; if you interleave backups, you will degrade restore speed.
Somebody will point out to them that it would be impossible to enforce a law like this, and it will be the last we hear of it.
Unless that "somebody" comes with a sufficiently large bribe, sorry, 'campaign contribution', they are likely to be ignored in favour of big businesses.
Hopefully common sense will win out, but I'm not too hopeful.
I don't have any original CDs in my car; I make copies to CD-R and use those. That way, they can get scratched, damaged or even stolen and I don't really care, I just make another copy. It's also easier to store a bundle of loose CDs than bulky cases in the car.
Never gives errors? Mine continually crashed when I thrashed the USB bus (e.g. copying files from a Compact Flash Card). Never did anything wrong in linux, even when doing the same actions, even if I did the same actions in VMWARE.
Now, these weren't your regular crashes; oh no, these consistantly fubar'd the registry making the entire box unbootable. The only fix I've found is to use a seperate USB card in the system to avoid crashes.
Yeah, saw it this morning. However, they have to be able to prove breaches of the licensees obligations (i.e. IBM & SGI) and let them know the details. If, after two months of advising IBM/SGI of the breach, they haven't fixed it, they can pull the plug.
Vague threats of "you're breaching our IP, but we're not telling you which part of our IP you're breaching" are not, IMHO, notifying them of a breach.
As for "all your code are belong to us" (regarding SCO's claim that XFS, JFS et al are "derivative works) really depends on the wording in the rest of the agreement.
That's actually a very good point. Get the basics right and add the popular features bit by bit. If you don't, you'll end up with 100s of half-assed features which don't work rather than a clean, stable app which may not do everything, but does what it does well.
I think it's an OS problem. Certainly, I haven't had an issue with games not working on switch-back under XP although ME gave me several problems switching back to games.
Slashdot Mirror
As for rack sizes, the 15K racks are about the same size as normal racks, but are slightly deeper. The system is not like a standard rack + servers, you buy a 15K as a single unit and add the system/IO boards.
Most people when they think of nuclear power worry about leaks, failures and "3 eyed fish" etc. However, those problems are fairly rare, but the issues regarding long-term nuclear waste storage are, in my opinion, more important. We're dealing with stuff that will be dangerous 100,000 years down the line.
In short, holding a piece of DU is fairly safe; breathing in DU dust isn't.
Having worked in a large corporation, budgets are one of the biggest roadblocks to this kind of thinking for two reasons: firstly, it's nigh on impossible to get extra funding even when it would give a benefit if you're out of budget and secondly, budgets lead to an end of year splurge to use up any remaining funds. "Accountants know the cost of everything and the value of nothing" sounds silly until you've worked in that kind of enviroment.
The main problem is that 90% of people don't know about filtering; they use whatever software their ISP provides. If they downloaded Mozilla or one of the POP3 filters, they'd get less spam but they just haven't heard of them.
Last I heard, Oracle still has raw disk as an option, mainly for performance reasons but also for parallel Oracle. With direct I/O or Quick I/O (Veritas), you can get near the performance of raw devices, but raw disk is still marginally faster.
Yes, more RAM is nice, but provided your OS doesn't take up most of it, you can do a fair bit in 128MB.
Backing up linux/Unix is a matter of backing up the filesystems with tar/dump/ufsdump/whatever to tape. Restoration involves booting from CD, reparitioning the disk and restoring from tape, installing a boot block and rebooting. If you're being picky, you shut down apps first and backup from single user mode, but 99% of companies get by fine with live backups (other than Oracle and DBs etc).
As far as I'm aware, the only way you can possibly back up a live NT system fully is to run disk mirroring and break off one mirror and back up that using some drive imaging. Restoration involves a complete restore of the entire filesystem, however.
The allies pretty much admitted that Al Qaeda wasn't in Iraq before the war, but they didn't make much of a noise about that because it served their purpose to have the public believe that Al Qaeda was in Iraq to bolster support for the war.
As for N Korea, Bush claiming it was part of his "Axis of Evil" didn't help. NK has now seen what has happened to one third of that (Iraq) and is now trying to make sure it isn't the next target. At this point, you can't really blame it for developing a nuclear deterrant.
Yes, clusters can do a job which is "good enough" to replace expensive mainframes, but there are some cases where they aren't good enough, especially banking where you have to be 100% confident that every transaction is logged correctly.
FWIW, the server name is transmitted in a standard HTTP/1.1 response so it's trivial to work out what kind of server something is running. As a simple test, run 'telnet [host] 80' and type 'GET / HTTP/1.1' and hit enter a few times. You'll get a response (usually an error saying invalid HTTP/1.1 request) which includes a server version.
Yup, it would be difficult. Especially at high altitude, as the plane undergoes severe decompression. There's a very good reason not to allow guns on aircraft, as firing a gun in an aircraft cabin will likely lead to an imminent crash.
I feel so dirty...
This has to be the best advantage of subscription; you get to beat the rush to view the site before the stampeding hordes of /.ers kill the poor server :)
Ah, if you use multiplexing then yes; OTOH, you can have multiple tape drives and have Netbackup stream one client to each tape drive and that will also speed up restores (it will be able to stream from tape rather than stream/seek). This is true for all backup products, not just Veritas; if you interleave backups, you will degrade restore speed.
On the rare occassions when I haven't had popup blocking on, I haven't seen an X10 ad for a couple of years.
Hopefully common sense will win out, but I'm not too hopeful.
I don't have any original CDs in my car; I make copies to CD-R and use those. That way, they can get scratched, damaged or even stolen and I don't really care, I just make another copy. It's also easier to store a bundle of loose CDs than bulky cases in the car.
Now, these weren't your regular crashes; oh no, these consistantly fubar'd the registry making the entire box unbootable. The only fix I've found is to use a seperate USB card in the system to avoid crashes.
Ultra 5 FCS date was December 1997 according to this.
Vague threats of "you're breaching our IP, but we're not telling you which part of our IP you're breaching" are not, IMHO, notifying them of a breach.
As for "all your code are belong to us" (regarding SCO's claim that XFS, JFS et al are "derivative works) really depends on the wording in the rest of the agreement.
That's actually a very good point. Get the basics right and add the popular features bit by bit. If you don't, you'll end up with 100s of half-assed features which don't work rather than a clean, stable app which may not do everything, but does what it does well.
I think it's an OS problem. Certainly, I haven't had an issue with games not working on switch-back under XP although ME gave me several problems switching back to games.