Posts by KingZongo
138 posts • joined Wednesday 19th August 2009 02:12 GMT
I always wondered...
...what happened to Jarvis Cocker. Never dreamed he worked for the Reg. Must be the mis-shapes.
Plus ca change
Funny how history repeats itself. Once upon a time enterprises had to rent compute time from IBM, who ran a nice little cloud operation on their mainframes. You wanted office apps? Sure, say hello to PROFS. Then some startups came along and said, this is nuts, let people run what they want without the high priests and priestesses getting involved. And we had an explosion in creativity. There are whole ecosystems built around what you can do with Excel and VBA.
Now Microsoft is turning back to the cloud. I seriously doubt that people are going to junk their investment in Office-on-PC and return to the old days.
Surely...
If this had happened in England, we would have had headlines like, Beak Goes Beserk As Burqa-Clad Beryl Beats Berkshire Beat.
Shakespeare 1, Voltaire 0.
Multiple writers? Over NFS? Gulp.
If my EMC rep tells me that they're selling a cloud-scale system and relying on NFS to keep multiple writers in sync, I'm running for the hills.
NFS locking = evil. Anyone who's spent days wrestling with lockd and statd knows what I mean.
Hopefully this is all a bad dream and there is no locking required because no two writers ever update the same data shard. As soon as locking gets involved, you are doomed.
Maybe this is announcement is what all that end of the world stuff was about.
RE: Sony needs to re-think? Are you kidding me?
The core of this is simple, but remains unaddressed in all the vitriol (in both directions). Does a consumer possess the right to do whatever they like with hardware that they have bought, or do they have to stick to the usage model that the manufacturer intended?
In the absence of a definitive ruling (which we will likely never get, because the impact would be _huge_), manufacturers will continue to try and get consumers to stick to the latter path without ever explicitly saying to consumers that they don't really "own" their own hardware.
Hardware mfrs that invest gazillions of $ in their platforms need to see return on their investment whilst simultaneously not blocking some of their sharpest users. Sony made what they thought was a sensible business decision to axe OtherOS and handled the followup badly. Conversely, GeoHot let his ego cloud his judgement and took the decision as a personal affront.
Would have been much easier for Sony to say "look everyone, we're axing OtherOS support. We're not going to support anyone who works around it or does something with their PS3 that we hadn't anticipated. But we won't prosecute them either. If they do anything illegal, for commercial gain, or that disrupts anyone who just wants to use their PS3 normally, we will come down on you like a ton of bricks". Simple, clear and fair.
Re: Respected??
Alright, I've got to ask: who makes 'good' servers these days? Where good is (my personal definition): built solidly, have field-replaceable components, don't slice your fingers up on bare metal when you pop the lid, made by people who understand that some of us have 19" racks (oo-er) and some have 23". Bonus points if they come in AC and DC power models.
Got some names? Dell used to be half decent (well, not compared to my much-loved and still-missed Sun 220Rs and E450s, but good enough), but not now.
Python?! Your kids will beg you to by them a real language
Any language where whitespace is significant is a no-no for children. Imagine explaining to an 8-year-old why their program didn't work because of the lack of invisible stuff at the start of a line.
I started my kids at six by helping them take an old computer apart and showing them what each part did, and explaining why computers are useless without instructions.
Around 8 they started playing with Scratch, getting the cutesy cat thing to move around. They learnt that there was this thing called Java under the covers, but that was it. The most important lesson you can teach a child about a computer is that there are _always_ things under the covers, and to really master computing requires a little bit of knowledge of what's going on down below.
Now they're 11 they watch as I program and sometimes roll their own. I don't fix their bugs, but I provide hints sometimes. We're also talking about how programmers need to manage the computer's resources, and code efficiency (at a very high level: eg "if you had to search your room for a sock, what would be the fastest way to do it?" leads on to a discussion of why some search algorithms are faster than others)
It may all be too late. My language of choice for quick jobs is Perl. I've probably scarred them for life.
Not a bad idea
Well, I salute anyone who tries to get kids to see under the hood. There are too many examples of black boxes in a kid's world that they should be educated on. Food doesn't just come out of a packet. Gas doesn't just come out of a pump. And computers don't do things by themselves.
Personally, I'd love to see something like a kit PS3. Let's say, you sell them the bits, no warranty, OtherOS support, at a knockdown price and you have the carrot of being able to play PS3 games at the end of your build. Would make a great Sunday afternoon project for the kids.
I know, never going to happen.
@variola
Since he was an evil man, clearly it would be Emacs (dons flamesuit at this point). Presumably on a 72 character green screen, to remind him of the number of virgins he was expecting to meet up with and green being the color associated with Islam.
It'll be really depressing if the West's nemesis has been plotting attacks using Word 97 or something like that. Visions of jihadi Clippy: "I see you're trying to blow up some infidels. Would you like help with that?"
Conglomerates
Not everyone subscribes to the (same) business fads as we have in the West. And in the developing/recently-developed world, brand loyalty is huge - you've never met an Apple fanboi like a Tata or a Godrej fanboi.
When I was young they used to advertise Godrej refridgerators as aspirational things - if you worked you ass off, you could aspire to own one. I think I'd rather have an ipad.
Irony of ironies
How deeply ironic that Nokia would be slayed by their inability to get an OS out of the door whilst Apple and Google are sitting pretty. Especially Apple: echoes of MacOS here. Remember the struggles they had with Taligent, Pink and so on before settling on a UNIX core. And didn't that turn out nicely?
Why is this news?
How is this news? There are lots of languages that co-exist quite happily. Parts of the US are bi-lingual English and Spanish already and one is not going to crowd out the other. And in India they have bazillions of languages, most vaguely similar (Hindi, Gujarati, Marathi etc) and some radically not (English) and they seem to rub along just fine.
As usual, the wrong approach
Why isn't anyone doing research into why spam is profitable? Somebody out there is clicking on those links, buying that viagra, and sending money to Lagos. Consistently remove the buyer from the market and useless spam disappears (leaving only phishing, which is easier to defend against).
Actually, I think it might be too late for open email. The future, to me, feels like web-of-trust email, where every sending address is blocked unles it is explcitly permitted by you. For example, 'accept mail from my Facebook buddies'.
It's rarely the OS
There should never be any surprises as far as servers and infrastructure boxes go: you knew what the app profile was like, how your setup would hit it and sized it appropriately so you should have ample notice when it's time to reach for a new one.
As far as desktops go, the real problem is not the hardware, but Windows. Everyone knows that Windows degrades over time, and speed is miraculously cured by a format + reinstall. As an enterprise It shop you just have to focus on making it fabulously easy to revert a user's PC to that virgin state with all their apps ready to go and you'll be fine.
Time was when my old machines would shuffle off to a new life running Linux. Sadly, Linux is so big these days that it's too painful to contemplate. OpenBSD is a very good alternative though, it's tiny, runs like a rocket and (unlike the various mini-Linuxes) has a full toolchain.
Von Erck - what??
OK so Raisley was a fool to go launching a bot attack. But scroll back to the source of all the angst - the falling out between Raisley and Von Erck. WTF was that about? They had a fight, Von Erck wanted some notion of revenge and set up sting? It's insane. Crazy folk.
Right in the long term, wrong in the short
Maritz' statements are partially true, but not the whole truth - pretty much standard for marketing I suppose.
He is correct that we are rapidly approaching the time when x86 hardware offers a type 1 hypervisor. That will obviate the need to run a host OS like Windows or Linux purely to offer a copntainer to a set of VMs.
He is incorrect that the VMs themselves will not need to run a guest OS. I can't see us getting away from that anytime soon. If I want to run some line of business app it'll have been developed for some OS and that OS will have to run as a guest in the container. Or is Maritz thinking that somehow developers will code to a set of OS APIs that VMWare will provide? That is a huge undertaking. Even IBM, with all their experience in OS development on S/390, never went that far. I'd like to see him try and sell that to his shareholders.
This is why applications should be anonymized
Much harder to discriminate when the adjudicator only has the facts of the case, not the fact that they are called Campbell/O'Campbell/Campbellowicz/Campbell-Patel. I think this should hold true not just of gov stuff like tax and benefits but college applications, jobs, the works.
Sad for us, right for Oracle
Much as it pains me to say it, this sort of laser-like focus on profit is why Oracle are buying Sun, and not the other way around. Oracle don't need to keep a bunch of HPC wizards in house, they have deep enough pockets to watch how the market matures and then co-opt or buy the new guys out. So if Supa-Dupa-Connect becomes the new I/O interconnect standard, they'll just support that, or buy it in, and they're done. No need to throw money at a bunch of projects only a handful of which will see the light of day.
I suspect that the Reg readership skews strongly to a sentimental feeling that Sun was a really really cool company with some truly ace talent. Which it was, for years. Unfortunately they pissed it all away and now we have Oracle making those hard decisions for them. Ugly, but there it is.
Very dull: plus ca change, and all that
Does any body care about this stuff any more? Once upon a time computers were big beasties that lived some way away from you, dear user, and you could only run what the men in white coats said you could. Then along came the PC and lo! you could run stuff whenever. The white be-coated men scoffed at your lack of a credible infrastructure but you persisted, until one day your CTO realized that maintaining a zillion PCS really was quite a pain and the cloud model might be worth looking at.
In short, we've been down this track before. For some people, and for some apps, a cloud model is fine. For other people, and other apps, it isn't. There's some grayness in between where a mixture of both might make sense, like HPC in the cloud and display on the edge devices. end of story. Nothing to see here folks.
Pretty straightforward here I think
The kindest explanation I can think of runs like this: he saw her, knew he had to have her and before you could say "bill and dave, fer chrissakes" he found an excuse to hire her (cmon, what part of her resume legitimately qualified her as a marketing person acting at the CEO level?). She took the money, then got cold feet when he started seriously hitting on her. It happens. That is why companies should institute a no-dating co-workers rule...
As far as @100113.1537's comment that it was Fiorina who did all the legwork of turning HP around, I have to disagree. Those were terrible, terrible years at HP. I can't pin the Itanium disaster on her but the Compaq merger, the splitting off of some of the best business units into separate companies, and the plain old-fashioned arrogance still grate on me 5 years after her departure.
Call me cynical but
Sourcefire need to chill.
(a) Suricata is barely off the drawing board
(b) It is unlikely to come to fruition for a couple of years
I suspect that Sourcefire are squealing because they stand to lose a ton of money in federal contracts if Suricata becomes the federal standard. But they need not worry too much, we've all seen this movie before. The federal government kickstarts adoption of a standard and the entrenched vendors squeal a lot, then realize that they need to play nice, since the Feds are a big customer, and then the market decisively dumps that standard in favor of the defacto market standard. OSI vs the IP stack, X.400/X.500 vs. SMTP and LDAP, IPv6 vs IPv4 spring to mind.
I predict that Sourcefire will realize that they make more money by working with Suricata than against it, co-opt some of the ideas, and sell "Suricata-Snort" back to the Feds. (Everyone else in the commercial market will carry on buying 'standard Snort', and eventually the Feds will get the picture, abandon Suricata-Snort and everything will be rosy again.)
An example of this behavior is when Microsoft realized that Exchange Server was unsaleable to the feds unless it could offer Defense Messaging System compliance, so they bolted on X.400 and X.500 support whilst standard Exchange moved rapidly to an an LDAP/SMTP system. Eventually the feds abandoned X.400/500 and went to the standard product. Note that it was not a complete waste of time - a lot of the expertise MS picked up in X.500 made its way into Active Directory.
You obviously never read Lyons then...
20 year old C code! Pah - Try 40 year old C code...as in the sterling examples in John Lyons source code tour of UNIX. Actually it's not nearly as painful as you might think. Probably because there was no room for cleverness in 4K and a single-threaded processor....
Everyone just needs to chill
We are long past the point of discussing whether this is an engineering design flaw or not. It really doesn't matter at this point. This is how it plays out at most corporates when this sort of stuff hits:
Lackey: Uh, sir, we have a problem with X
CEO: You have a problem with eggs? Change your diet. I like salmon. Do you like salmon? And caviar. Do you like caviar?-
Lackey: --No sir, X. X. We have a problem with X. The <antenna is shagged|the oil rig blew up|our drugs cause peoples noses to turn blue|....>
CEO: Oh.
CEO: Who knows about it?
Lackey: 3 million people. And the lady from the New York Times that you made a pass at last Christmas
CEO: <closes door>. Sit down, Smithers.
and then, one of
CEO: Deny everything. Tell them they're doing it wrong.
CEO: This is terrible. We need to do a recall. Oh, and is there a children's hospital we can sponsor after this blows over? Or maybe some puppies?
CEO: Deny everything. Tell them they're doing it wrong. <Then issue a patch|pump the hole full of golf balls|tell them blue is the new black>
I think it pretty clear that Apple chose #3 and as a result are going to be the poster-child example of how not to handle a product flaw. Your best best is to over-communicate ("we've heard reports of a problem, we're not sure how widespread it is, we're working with our partners to find out, whatever it is, we'll solve it for our customers") Then you go and lean on your customer facing folks to supply a workaround ("pick up your free bumpers at any Apple store") whilst you lean on your engineers to fix it (move the gap to the top of the phone, encase it in plastic, whatever). Then you start advertising again, heavily.
Sigh. They should pay me.
In defence of Mike
I'm with Mike on this one. Quite frankly, he is well within his rights to be cautious. We're talking about the company's crown jewels here - 10 petabytes of data. With that much on the line, he would be a fool to go out on a limb with a little vendor when the big guys give him what he wants -- even if he has to pay extra for it. Bear in mind that at this scale he is not just paying for fancy disks but also the kind of support that gets top-notch engineering asses on planes at the drop of a hat. All of which keeps the CFO happy and Mike in employment.
Sure the other guys can deliver more for less. What they have to do is to get into some lower-tier accounts, and pull out all the stops to make sure they execute flawlessly. Once they have a bunch of those under their belt they can start going up the food chain to the fortune 500s. At this point they either get co-opted (everyone adopts their technology), bought out, or they succeed by themselves.
Related, but only in the way that a gocart and a car are
Sybase and SQL Server come from the same bloodline, and were very close up until about version 4 (ie about 15 years ago) but they long, looooong since diverged, even in the engine. Sybase was always a damn fine database though. Used to be very popular in financial markets, at least until the Oracle dudes came knocking with their deep pockets and deeper discounts.
Welcome to the world of cellphone as d*** extension
Hmm. Not to be outdone by the President's Sectera Edge, Sarko now has an incredibly ugly but no doubt nicely encrypted phone. Presumably every G8 leader will soon be touting his/her own crypto-phone as they vie for geek superiority. Oy veh.
First of all, this is barely news - fair enough for the Reg to report it but really every high-level govt employee should get a crypto-phone just like they should get an encrypted laptop. The real problem is people leaving said devices on the train home...or deciding that it'll all be ok if they just call the Minister on their iPhone because the crytpto-phone is too tedious to use.
Too little, too late
Maybe I'm not representative (I liked elm, for god's sake) but I always thought that t-bird was fat with cruft and really missed the point. I think it's too late now; the days of a separate email client for consumers are over. Remarkably, the two dominant web UIs -- hotmail and gmail -- are absolutely rubbsh still, mind. Gmail's interface when you have a coupla hundred email threads is like an explosion at at a newspaper factory.
Nice one Sage.
Hats off to Sage who must have spent many sleepless nights worrying if Microsoft were going to come in and eat their lunch. A rare example of the superior product (and the _infinitely_ superior tech support) winning through.
@david W - yes, it's true
It's true, though I suspect it is more of a comfort law (makes legislators feel better about themselves) than useful for its original intent, which was to stop terrorism. See http://security.homeoffice.gov.uk/ripa/encryption/faqs/
The law is, I think, a response to those ticking-bomb scenarios. So, there's a bomb on a bus someplace, and MI5 have the guy, but wait! - somehow they know that his laptop has the map, but it's all encrypted! We'll force him to give us the password. This might work for a typical citizen but terrorists aren't going to be persuaded by the threat of 2 years in the clink vs. pulling off their mission, meeting the 72 virgins (or is it raisins? Either way, it always seemed pervy to me) and so on.
Offshoring isn't automatically bad..and another classic COLT story
Gotta say, there are an awful lot of very bright Indians out there. So don't assume that just because colt are outsourcing to bangalore or wherever that they're going to get idiots. And let's face it, given colt's record things might actually improve.
My favorite COLT story comes from a friend who had to support an app running on colt's servers. They called up one day in a tiz cos the servers had dropped off the network. Obviously an outrageous outage. Could he find out wtf was going on?
Few hours later, turns out that the techs in the CO had decided the servers couldn't be very importasnt cos no one ever visited them so they pulled them out of the racks and stuffed them in a Transit van. Stellar.
The only thing new in software...is that there's nothing new in software
Ahh, Google reach Stage Four of the rise and fall of great IT companies. The stages being Startup, Innovation, Dominance, Hubris and Death. A marker of hubris is to reinvent the wheel "only this time do it better". Virtualization is a classic, recurrent theme here and almost invariably ends up dying in the Tarpits Of Performance, a strange place where the promise of new hardware dangles acceptable ops/sec in front of the designer but never quite delivers.
In this case the tarpits have claimed yet another scalp. Face it boys and girls, nothing runs as fast as code for the metal you are running it on. Either learn to live with it, throw a lot of hardware at it (not PC hardware, I'm talking the z/VM approach) or spread the load over nodes by cheating (X window terminals, arise)
It'd be a brave MIS manager who bought SGI
I used SGI kit for HPC back when they were the old, purple-beastie SGI and loved it. But it'd be a brave purchaser who bought their kit to run ERP or middleware.
- Standard hardware?: err, no, not remotely. NUMA I can live with (lots of people ran Oracle on Sequent Dynix +NUMA after all). But the use of Itanium would scare my Forbes-reading colleagues
- Viable vendor?: Hmm, maybe, but their commitment to the data center is yet to be proven. They're certainly no Dell or Sun here.
- Cheap? Hell no.
As for the comments that Java might be a good HPC language, I'm skeptical. For all their faults, Fortran and C offer access to really whopping amounts of memory _in a deterministic fashion_ which you need when you are modeling genomes and suchlike. I'd love to know how they get Java to access a couple of Tb of RAM and not occasionally think that it ought to do a stop-the-world garbage collection. -Xmx=1000000Gb --no-garbage-collection maybe ? Would make for a very interesting article.
money talks
It's actually quite simple to solve this problem. Your Internet connection fees need to be charged by the byte, with itemized bills going out to users that lists the sites/IPs they have connected to. Nothing will make an end user update their software more quickly than being asked to pay for a few gigs worth of botnet traffic; it'd be like discovering your kid has been dialling premium-rate phone numbers and likewise would be shut down immediately. The savvy ISP would then sell a cheap subscription services where they would install patches and updates to the end users. Profit all round, problem solved.
I'd also point out that it would have the delicious side effect of putting all those advertising popups out of business overnight as people would not want to pay for a zillion Netflix pop-unders from casalemedia (etc.). As well as pushing the web to become leaner and lighter, something that is long overdue.
Hinges on the definition of ownership
Like many recent digital advances this hinges on the definition of ownership. Do I own my cellphone or merely lease it from the provider for 2 years? If I own it, competition law ought to say it's mine to do what the hell I like. Bored with AT&T? Fine, go stick a Verizon SIM in the back and move on. If I lease it, on the other hand, I am really stuck with the carrier and all the law can do is say that after the two year period I should be able to trivially unlock my device and go to whoever I like.
Personally I would rather pay the full price for a handset and be able to do what the hell I like with it rather than have the price obscured in a two year contract. Any kind of cross-subsidy or price occlusion is always to the detriment of the customer. (Exhibit A is the US healthcare system, but let's not even go there today.....)
Smells fishy
I find it hard to believe that someone could enter the US on a temporary visa and then, on the strength of attesting in an online form that they made $180,000, get an unsecured credit card with a $25k limit. The card issuer would have run a credit score check and found no history and that alone would have been grounds for a rejection. There's more to this story than meets the eye.
They've seen the future, and it doesn't include them
I'm not sure why any incumbent carrier/cableco would want to sink $millions into a nationwide buildout at this time.
Even if the taxpayer funds the capex the opex is hideous and coupled with the fact that no-one under 25 has a landline, that broadband experience will be available over cellular networks by the time the thing is built, and that the real profit flows to the content provider (google) not the carrier, I can see why they would want to pass.
I'm not entirely sure what their plan is for the next ten years, but I think we'll see carriers offer fantastic services in large population centers and ignore the rest of the US.
