``While OSI is relatively simpler and more clear cut'' --- the model, perhaps. The protocols were horrific. Yes, I've done implementation work with them. All hail Steve Kille and Julian Onions, but just think what they'd have achieved if they hadn't tried to make OSI implementable!
Forgetting about facts and looking just at public perception, if people (and presumably, if they're in law school, they're not drooling idiots, either) regard the move from XP to OSX as less scary than the move from XP to Vista, that's a big problem for Microsoft. It's a horrible problem for pitching new releases: make it sound the same and no-one has a reason to take it, make it sound like a major shift and people may as well evaluate other vendors' products if they're going to have to make a large change anyway.
`` It'd be a pretty hard sell to explain to 11 people who have at best a passing familiarity with computers how spoofing and viruses introduce an element of doubt into who has what IP address on a filesharing network. ''
And it'd be a pointless sell, because in a civil court reasonable doubt is not enough: it's about the balance on probabilities. It sounds like she and her lawyer thought they were in a criminal court.
``Thunderbird is far more extensible and has quite a few features Apple's client lacks, like good IMAP folder management and Bayesian filtering.''
Huh? Mail.app is the first GUI client I've found tolerable (Multics read_mail 1983-86, MH and nmh 1986-1999 with a brief flirtation with RMAIL in there somewhere, then mutt when I needed IMAP4 support, now Mail.app since I've drunk the Jobs Koolaid). It has Bayesian filtering --- its junk filtering is both Bayesian and ``pay attention to ISP headers''. I'm not sure what you mean about IMAP folder management, but I have several hundred IMAP folders, in deep hierarchies, sometimes with messages in the intermediate nodes, and it all seems to work OK for me against Cyrus.
From The Guardian (UK's main soft left newspaper): How Russia lost the moon
Sergei Khrushchev: The Soviets squandered the lead in the space race that Sputnik gave them, despite my father's efforts. You can read it at http://www.guardian.co.uk/comment/story/0,,2181418,00.html.
The Germans under Hitler, for example, were quite good at pushing the technological envelope in some areas.
But by dismissing quantum physics as `Jewish Science' Hitler wrote his own death warrant. There is no alternative history of WW2 where Germany wins: it's just that Hiroshima would be a small town in western Japan famous for a floating gate, and Berlin would be famous as the scene of the first active nuclear explosion.
Protocols mandate close synch of UTC. If they required clock-on-the-wall time to be synchronized then they wouldn't work across time zone boundaries. I manage a network with staff in six timezones, between UTC-8 and UTC+9, some of them with non-integer offsets, with (from memory) four different sets of DST rules, including one (Japan) that doesn't do daylight saving. I know about the timezones because I'm a clock geek sad enough to know the difference between UTC and GMT. But I don't need my systems to understand it (aside from shouting at bloody Apple for thinking BST, UTC+1, is called BDT). That's because everything that cares about timing (Clearcase, WebDAV, SVN, NFS, Make...) works in UTC.
Solaris geek time. Processor sets are a bit brute force, unless you have a very specialised workload. Placing distinct application sets into distinct projects, and then allocating those projects cpu shares for the fair shares scheduler to work with is a more general solution. Sometimes you do it to lump aspects of the same workload together, so that two workloads can co-exist reasonably. Sometimes you do it to keep aspects of the same workload together, so that one part can't swamp another. We did some judicious tuning of an old E450 that provides Cyrus mail for ~600 people, and found that keeping imapd and lmtpd in distinct projects, with nscd and bind in a third, allowed load averages of 150+ (no typo) and 95% CPU utilisation to nonetheless deliver decent response.
``Basically chroot was an early attempt at virtualization. It allowed one to keep servers separated and contained, ''
chroot() appears in seventh edition (http://minnie.tuhs.org/UnixTree/V7/usr/src/libc/sys/chroot.s.html), released in January 1979. It's written in PDP11 assembler. It's a hell of a stretch to argue it was meant for virtualisation, server containment or whatever use it might be put to today.
Although to be fair, the same was said XP at the time: for people coming off the 95/98/ME train, their voodoo didn't work. So they promoted the idea that XP was a step backwards in order to extend the life of their voodoo. Non-geeks don't like change, and slightly-geeks don't want to have to learn a new load of voodoo to impress not-geeks.
It's not the same story once bankruptcy is involved. If SCO had to pay a 95% royalty to Novell, then SCO owe money to Novell. Novell can then just get in line if SCO go bankrupt, and get (potentially) cents in the dollar, if and when there's a distribution by the bankruptcy court. If SCO collected the money on behalf of Novell and then retained a 5% commission, then Novell can take action independent of the bankruptcy court to protect their property. The position I outline, which is what is in the contracts, places Novell in a fantastically stronger position.
Novell licensed its Unix to SCO and asked for a ~95% royalty on sales.
No. Novell licensed its Unix to SCO, and offered SCO a 5% handling charge on collecting Novell's royalties. What's the difference, you ask? The difference is that such money from the licenses that SCO hold is Novell's in equity, not as a creditor. Novell are like the people who leased SCO a photocopier: it's their's, and they can take it without reference to other creditors.
If you steal a car and then go bankrupt, your creditors can't fight over who gets the car: it's not yours, so it's not an asset in bankruptcy, and it goes to whoever holds title. Same here.
Of course, in a country with a sensible data protection regime, forwarding personally identifiable information to a weakly-protected gmail account would be a non-no in and of itself, One of the problems with the US's absolute lack of constraints on companies' use of personal data is that the casual mailing of SSNs can go on, and management have no reason to deal with it. In europe, that sort of stuff is locked down into HR department systems.
DH is in BQP along with RSA, though, so would presumably be equally at risk. Elliptic curves aren't as well studied (at least in the open community). Everything else is pretty speculative, or equivalent to existing problems.
A lot of people lose sight of this, but there's no reason why if you need "only" 2TB, that you _buy_ only 2TB. There's no reason why a "2TB array" can't be 30*500G drives from which you only access the first ~130G and ignore the rest (as you say, storage is so cheap the "waste" is irrelevant).
Power, space, heat, MTBF, complexity. I'm using some Pillar equipment that short-strokes SATA drives but makes the residue available at a lower QoS, which works well, but I can make use of the residual space. I've done what you propose in the past, but it's a lot more expensive over the whole life of the boxes than you make out.
``I think NetApp is also worried about Apple. ''
Apple don't have the channel or service partners today. It would take them years to develop those, and the enterprise storage market is a scary place to learn those lessons. How would Apple offer 24x7 2hr engineer on site service? How would they convince their existing channel to get involved in what are often twelve-month sales cycles even for established kit?
I'm an Auspex customer from back before NetApp even existed, and buying that (14GB! $200K! ah, the early 90s) was an eighteen month exercise. I'm currently getting my big black NAS fix from Pillar (great company, good box): that was a two year sales cycle from initial approach to final acceptance. In both cases, the vendor is fronting up resource, loan machines and development effort, and the prospect can walk away at any time. Channel willing to do that isn't Apple's forte.
NetApp had all these problems when they started up --- I think the first one we bought had a spares kit and a good luck note, and I bought it very much on the strength all all the ex-Auspex guys (and Guys). But times were different: they had a compelling product, the market was growing explosively, Auspex were busy blowing their feet off with automatic weapons and Sun weren't seriously in storage. Today, a new entrant into NAS has to contend with NetApp, EMC and Sun at tier one, then Pillar, 3Par, BlueArc and all the rest at tier two. Apple's branding is an actual negative in that space, and there's a lot of competition and not a lot of loose margin. Hard.
2TB is easy, but 2TB with decent performance is rather harder. 2TB with decent performance and good post-power-off performance is harder still. Back in the day, 2TB would be spread over 60 or more 7200rpm spindles, whereas today it's on half a dozen 7200rpm spindles. You've got the capacity, but you've got a tenth of the ops/sec. This might well not matter to you, if you're doing data warehousing or serving home directories, or it might be very important indeed if you're doing a lot of OLTP.
NetApp still have a lot of margin in their product, and if you want storage that one vendor will look after, they're pretty damn good. The thing to remember is that storage is so cheap now that in a project that needs 10TB, the price of the 10TB is in the noise floor compared to the price of the people, integration, application development, user training, etc, etc. So shaving 20% of the storage costs is perhaps 2% off the project, and that might be worth it for the fact that NetApp kit just works.
I happen to have recently bought 40TB from one of their competitors (Pillar), and I'm in the process of building a 30TB ZFS farm on a mix of Solaris and Dot Hill, but there's a NetApp in the corner that works extremely well.
Video recorders didn't distribute copies of films before they were released.
And yet, sat at the end of fast links both at home and in the office, with terrabytes of disk space, aged 42 I'm sat in a darkened room with an ice cream in my hands once a week.
Does fair use cover posting the ending of a new movie the day before it is released?
So what? I know how The Wickerman ends, and I know how The Dambusters ends. It's not hard to guess how most new films will end. I'm pretty certain Macbeth won't be changing its end anytime soon, but I've seen a dozen or more productions of it in theatres, too.
Do you think the movie studios are going to keep making movies when everyone is just downloading and watching them at home?
Which is the crap that people spouted when video recorders were about to bring Hollywood to its knees in the 70s. Cinema attendances are healthy, because people like going to cinemas. It's not like spending time at home. If people drift away from cinemas, the villain is not downloading, the villain is either (a) the films or (b) the cinema-going experience. Both are quite clearly in the scope of Hollywood itself to fix.,
Over the past few weeks, I've seen cleaned digital prints of Henry V, The Dambusters and The Wickerman in cinemas. They were close to full. I couldn't see Goldfinger, because it has sold out before I got there. These films are between thirty and sixty years old, but still pull an audience: big screen, good sound, good film, nice place to be.
The business model might change. Nicole and Brad might not get as much money. But cinemas will be with us as long as teenagers want to get out of the house.
What if it were a house and the builder put these limitations on it?
``Restrictive Covenants'' (http://en.wikipedia.org/wiki/Restrictive_covenant), are perfectly legal. I'm subject to English property law (which is different to Scottish, never made US) property law, but Wikipedia says they are similar for Americans.
The usual one is a restriction on a house built on an estate (subdivision in US terms? I dunno) preventing change of use even if you manage to get planning permission. There's a clause in the deeds for my previous house preventing using it as an off-license (liquor store) even if I can get planning consent and a license, put in by the brewery that originally owned the land in the 1930s.
It's routine for houses to have restrictive covenants which prevent you from being a pain to your neighbours: someone I work with has a house with a restrictive covenant covering parking caravans outside.
They're tricky to enforce, and you can apply to courts to have them lifted. But the concept is basically sound.
For advanced English readers, you can explore how they interwork with the Leasehold Reform Act for houses in two Birmingham suburbs, Port Sunlight and a handful of other locations which historically used leases to achieve the same effect. As a resident of Bournville (yes, the chocolate) things get very complex...
You're taking the wrong lesson from Minitel. The real lesson is that pan-national will always triumph over local. See how far OSI got, even in the countries that mandated it (such as, of course, Japan).
So long as your house doesn't burn down. And you don't get burgled. And you don't live somewhere prone to flood or hurricane. I'd merrily pay that sort of money for that sort of storage, because moving portable drives offsite is getting tedious. With my office hat on, the fibre pull completes Tuesday for 2Gbps to a datacentre in the next city to replicate 40TB to. Yeah, I have tapes: but moving them by hand is tedious. Same idea, smaller bill.
For anyone wondering why an hour my local cinema complex rarely has a empty theatre and its a good idea to get there 20 minutes early.
Which points up the idiocy of the claims that `piracy' is killing cinemas: Cinema attendances are healthier than they've been since the early 1970s. There are about 35 screens within a mile of the centre of Birmingham, and a further fifty to a hundred within five miles. In the late 1970s cinemas were closing, or being converted into bingo halls and concert venues. Every cinema I went to as a child in the sixties and seventies had gone by 1985, and today's cinemas are all new build. If cinema's in such trouble, why did a new 12 screen open a couple of years ago only two hundred yards from an existing 10 screen?
I know full well I could have got a 720x576 with 5.1 sound Xvid file off the internet days earlier than the films release. Why are they telling us to goto the cinema when I'm there, I'm in the cinema because I can't get a 20ft screen with and incredible sound system at home you've won just stop lieing to me.
I shout that complaint at the screen (silently) every time I see that silly advert. Stop hassling me: I've paid. Hassle the people who haven't. Stop treating your paying customers, in my case regular paying customers, as criminals.
Here outside London a trip to the cinema costs about five pounds (five quid, a fiver). You can typically buy an `as many films as you want' pass for about twelve quid a month, but tied to one cinema. Which works if you see all your films at that cinema, but is a less good deal if you sometimes go to an arthouse or whatever. It superficially pays back at the third film, but there are a variety of offers (Tuesdays are typically only three pounds, before six o'clock most days ditto, family tickets which essentially give the parents the child reduction, etc, etc) which reduce the average cost. If your mobile phone is from the operator `Orange' then they have a huge nationwide promotion, `Orange Wednesday', which offers two-for-one cinema tickets to holders of their phones.
Tonight I'm going to see a digital projection of a clean transfer of `Goldfinger'. It's priced at £2.50.
Slashdot Mirror
``While OSI is relatively simpler and more clear cut'' --- the model, perhaps. The protocols were horrific. Yes, I've done implementation work with them. All hail Steve Kille and Julian Onions, but just think what they'd have achieved if they hadn't tried to make OSI implementable!
ian
`` It'd be a pretty hard sell to explain to 11 people who have at best a passing familiarity with computers how spoofing and viruses introduce an element of doubt into who has what IP address on a filesharing network. '' And it'd be a pointless sell, because in a civil court reasonable doubt is not enough: it's about the balance on probabilities. It sounds like she and her lawyer thought they were in a criminal court.
``Thunderbird is far more extensible and has quite a few features Apple's client lacks, like good IMAP folder management and Bayesian filtering.'' Huh? Mail.app is the first GUI client I've found tolerable (Multics read_mail 1983-86, MH and nmh 1986-1999 with a brief flirtation with RMAIL in there somewhere, then mutt when I needed IMAP4 support, now Mail.app since I've drunk the Jobs Koolaid). It has Bayesian filtering --- its junk filtering is both Bayesian and ``pay attention to ISP headers''. I'm not sure what you mean about IMAP folder management, but I have several hundred IMAP folders, in deep hierarchies, sometimes with messages in the intermediate nodes, and it all seems to work OK for me against Cyrus.
From The Guardian (UK's main soft left newspaper): How Russia lost the moon Sergei Khrushchev: The Soviets squandered the lead in the space race that Sputnik gave them, despite my father's efforts. You can read it at http://www.guardian.co.uk/comment/story/0,,2181418,00.html.
Protocols mandate close synch of UTC. If they required clock-on-the-wall time to be synchronized then they wouldn't work across time zone boundaries. I manage a network with staff in six timezones, between UTC-8 and UTC+9, some of them with non-integer offsets, with (from memory) four different sets of DST rules, including one (Japan) that doesn't do daylight saving. I know about the timezones because I'm a clock geek sad enough to know the difference between UTC and GMT. But I don't need my systems to understand it (aside from shouting at bloody Apple for thinking BST, UTC+1, is called BDT). That's because everything that cares about timing (Clearcase, WebDAV, SVN, NFS, Make...) works in UTC.
Solaris geek time. Processor sets are a bit brute force, unless you have a very specialised workload. Placing distinct application sets into distinct projects, and then allocating those projects cpu shares for the fair shares scheduler to work with is a more general solution. Sometimes you do it to lump aspects of the same workload together, so that two workloads can co-exist reasonably. Sometimes you do it to keep aspects of the same workload together, so that one part can't swamp another. We did some judicious tuning of an old E450 that provides Cyrus mail for ~600 people, and found that keeping imapd and lmtpd in distinct projects, with nscd and bind in a third, allowed load averages of 150+ (no typo) and 95% CPU utilisation to nonetheless deliver decent response.
``Basically chroot was an early attempt at virtualization. It allowed one to keep servers separated and contained, '' chroot() appears in seventh edition (http://minnie.tuhs.org/UnixTree/V7/usr/src/libc/sys/chroot.s.html), released in January 1979. It's written in PDP11 assembler. It's a hell of a stretch to argue it was meant for virtualisation, server containment or whatever use it might be put to today.
Although to be fair, the same was said XP at the time: for people coming off the 95/98/ME train, their voodoo didn't work. So they promoted the idea that XP was a step backwards in order to extend the life of their voodoo. Non-geeks don't like change, and slightly-geeks don't want to have to learn a new load of voodoo to impress not-geeks.
It's not the same story once bankruptcy is involved. If SCO had to pay a 95% royalty to Novell, then SCO owe money to Novell. Novell can then just get in line if SCO go bankrupt, and get (potentially) cents in the dollar, if and when there's a distribution by the bankruptcy court. If SCO collected the money on behalf of Novell and then retained a 5% commission, then Novell can take action independent of the bankruptcy court to protect their property. The position I outline, which is what is in the contracts, places Novell in a fantastically stronger position.
If you steal a car and then go bankrupt, your creditors can't fight over who gets the car: it's not yours, so it's not an asset in bankruptcy, and it goes to whoever holds title. Same here.
ian
Of course, in a country with a sensible data protection regime, forwarding personally identifiable information to a weakly-protected gmail account would be a non-no in and of itself, One of the problems with the US's absolute lack of constraints on companies' use of personal data is that the casual mailing of SSNs can go on, and management have no reason to deal with it. In europe, that sort of stuff is locked down into HR department systems.
DH is in BQP along with RSA, though, so would presumably be equally at risk. Elliptic curves aren't as well studied (at least in the open community). Everything else is pretty speculative, or equivalent to existing problems.
I'm an Auspex customer from back before NetApp even existed, and buying that (14GB! $200K! ah, the early 90s) was an eighteen month exercise. I'm currently getting my big black NAS fix from Pillar (great company, good box): that was a two year sales cycle from initial approach to final acceptance. In both cases, the vendor is fronting up resource, loan machines and development effort, and the prospect can walk away at any time. Channel willing to do that isn't Apple's forte.
NetApp had all these problems when they started up --- I think the first one we bought had a spares kit and a good luck note, and I bought it very much on the strength all all the ex-Auspex guys (and Guys). But times were different: they had a compelling product, the market was growing explosively, Auspex were busy blowing their feet off with automatic weapons and Sun weren't seriously in storage. Today, a new entrant into NAS has to contend with NetApp, EMC and Sun at tier one, then Pillar, 3Par, BlueArc and all the rest at tier two. Apple's branding is an actual negative in that space, and there's a lot of competition and not a lot of loose margin. Hard.
ian
2TB is easy, but 2TB with decent performance is rather harder. 2TB with decent performance and good post-power-off performance is harder still. Back in the day, 2TB would be spread over 60 or more 7200rpm spindles, whereas today it's on half a dozen 7200rpm spindles. You've got the capacity, but you've got a tenth of the ops/sec. This might well not matter to you, if you're doing data warehousing or serving home directories, or it might be very important indeed if you're doing a lot of OLTP. NetApp still have a lot of margin in their product, and if you want storage that one vendor will look after, they're pretty damn good. The thing to remember is that storage is so cheap now that in a project that needs 10TB, the price of the 10TB is in the noise floor compared to the price of the people, integration, application development, user training, etc, etc. So shaving 20% of the storage costs is perhaps 2% off the project, and that might be worth it for the fact that NetApp kit just works. I happen to have recently bought 40TB from one of their competitors (Pillar), and I'm in the process of building a 30TB ZFS farm on a mix of Solaris and Dot Hill, but there's a NetApp in the corner that works extremely well.
Over the past few weeks, I've seen cleaned digital prints of Henry V, The Dambusters and The Wickerman in cinemas. They were close to full. I couldn't see Goldfinger, because it has sold out before I got there. These films are between thirty and sixty years old, but still pull an audience: big screen, good sound, good film, nice place to be.
The business model might change. Nicole and Brad might not get as much money. But cinemas will be with us as long as teenagers want to get out of the house.
ian
The usual one is a restriction on a house built on an estate (subdivision in US terms? I dunno) preventing change of use even if you manage to get planning permission. There's a clause in the deeds for my previous house preventing using it as an off-license (liquor store) even if I can get planning consent and a license, put in by the brewery that originally owned the land in the 1930s.
It's routine for houses to have restrictive covenants which prevent you from being a pain to your neighbours: someone I work with has a house with a restrictive covenant covering parking caravans outside.
They're tricky to enforce, and you can apply to courts to have them lifted. But the concept is basically sound.
For advanced English readers, you can explore how they interwork with the Leasehold Reform Act for houses in two Birmingham suburbs, Port Sunlight and a handful of other locations which historically used leases to achieve the same effect. As a resident of Bournville (yes, the chocolate) things get very complex...
ian
ian
So long as your house doesn't burn down. And you don't get burgled. And you don't live somewhere prone to flood or hurricane. I'd merrily pay that sort of money for that sort of storage, because moving portable drives offsite is getting tedious. With my office hat on, the fibre pull completes Tuesday for 2Gbps to a datacentre in the next city to replicate 40TB to. Yeah, I have tapes: but moving them by hand is tedious. Same idea, smaller bill.
Here outside London a trip to the cinema costs about five pounds (five quid, a fiver). You can typically buy an `as many films as you want' pass for about twelve quid a month, but tied to one cinema. Which works if you see all your films at that cinema, but is a less good deal if you sometimes go to an arthouse or whatever. It superficially pays back at the third film, but there are a variety of offers (Tuesdays are typically only three pounds, before six o'clock most days ditto, family tickets which essentially give the parents the child reduction, etc, etc) which reduce the average cost. If your mobile phone is from the operator `Orange' then they have a huge nationwide promotion, `Orange Wednesday', which offers two-for-one cinema tickets to holders of their phones.
Tonight I'm going to see a digital projection of a clean transfer of `Goldfinger'. It's priced at £2.50.