Re: What’s the compute for?
Or you are uploading processed data that's a fraction of the sized of the ingested data, and you built the solution in AWS by mistake.
163 posts • joined 24 Feb 2015
Don't expect computer scientists to come to the rescue. From what I've seen, the way coding is taught starts with conceptual elegance, and does not always end with operational efficiency.
Take objects and recursion. Conceptually clean, operationally voracious. Both are abused like catholic altar boys: much code is neither elegant nor efficient. And so when it comes time to optimize, it may be that 'you can't get there from here.'
NAT is not evil, but it will eat itself. My modest home router handles about 40k packets a second. Soon it will be 80k Eventually it will be a 800k. NATting that would be a baroque and bloody-minded computational overhead. Why delay the inevitable?
I'm running dual stack on Deutsche Telekom. I don't know whether I have this lightweight thingy. Fixed IPv4s work fine locally. The additional effort is zero.
Just switched from 16/2.5 (measured 14/2.5 for years, but had fallen to 7/1) to 50/10 (measured 70/20).
To repeat, I switched from 7Mbps to 70Mbps down, measured. *I generally can't tell the difference*. That's because I had impeccable latency before, sitting, as I am, on a global hub.
My real and not very original point is that downlink speeds don't mean a cold damn. Be careful what you wish for. Goodheart's Law says that if you fetishize downlink speeds, or allow politicians to do so, you will end up with a shit service. Our job as techies is to encourage enlightened debate about the real requirement here.
I've used VDI for years, chiefly for manageability, and have always been amazed at the resources I've had to throw at it to make it work okay. For instance 1:
"[Gartner] said VDI works nicely with latency of between 100 and 180 milliseconds, but at 200 milliseconds your users will be burning you in effigy."
Like Gartner itself, anything over 50ms is for occasional use only.
For instance 2:
Is your VM rubbish? Low latency will not help if your VM is rubbish. Can be hard to test all remote.
For instance 3:
Windows: It's a bulimic pig (with apologies to bulimic pigs everywhere), puking great mounds of throwaway I/O. Sheer passive aggression by Microsoft is the laziest hypothesis.
So is VDI cheap? Lavish storage and networking are important clues. Is it unavoidable? Almost never. Is it nice to have? Yes.
No, this is a lawyerly tactic. These suits are driven by the corporate equivalent of ambulance chasers. They crawl the data for stocks that have fallen, advertise their services, and put boilerplate class action suits together by the dozen. The objective is to be just credible enough to force the ostensible victim to settle to make them go away. The class action clients get mostly nothing out of it, and you could argue they are the real mark in this game.
Pond scum is a glowing compliment for this grotty crevice of the legal profession. They aspire to the exquisite horror of the lamprey, but most are mere parasitic flukes and annelids.
The SEC does a perfectly good job of holding listed companies to account for full disclosure and is far more feared than these craven guinea worms.
Given the value of slots in the main hubs, it wasn't unreasonable for Airbus to bet that bigger craft would help airlines. Airbus may also have bet on slot allocation reform, where airlines would be made actually to rent slots, and bid for them properly, instead of squatting. If this happens one day, then the A380 may end up being the right plane at the wrong time.
You would do worse than have a think about how the FSA used to work. Its premise was that transparency made for better retail decisions. The rest kind of takes care of itself.
Transparency is an obvious target for regulators because there is a clear incentive for sellers to obfuscate, which constitutes market failure.
So for a set of service providers, sample and publish these (familiar to networking folk) indicators, for urban and rural, consumer and business, wired and wireless connections, and try to educate people on what they mean:
If a 5G or fibre fetish makes these numbers better, people will buy them. If they don't, they won't. The end.
Our experience has been that you can't let kids watch online video unattended. The algo will just keep choosing new shit to watch, which gets shitter as the chain of association gets longer. And little kids will watch anything.
Later on in the growing up process, a not inconsiderable problem is an education system that is neither equipped nor designed to teach critical thinking. If you watch random YouTube, then you are probably neither designed nor equipped to teach this to your kids, either.
Nothing changes. Here in Germany, you can get a rebate for buying a diesel, as long as you can find an old one to scrap ('old' means up to 2009). And you get a lifetime diesel fuel discount given that diesel is taxed much more lightly than petrol. Since everyone on Germany works for auto manufacturers, or should, it's kind of win-win.
For the enterpreneurs who export the 'scrapped' cars to north Africa and eastern Europe, it's win-win-win!
Admittedly, every time you fill up your subsidized diesel, a kitten dies. Most people accept this, until it's their kitten. A dead kitten rebate would go a long way to compensating people for this unavoidable loss.
On the positive side, finding a charging station is the least of our worries. With all these new cars, there isn't anywhere to park in the first place.
Same questions as for enterprise drives.
2. Rebuild times. Halve the nominal throughout number and divide it into the capacity. For parity RAID, divide by three. You're looking at a worst case of 36 hours during which you may need to engage the services of a priest. Do not try your own voodoo!
3. Vibration. Tolerances seem to have widened over the years. Consumer drives are shakier and noisier than they once were. This is a bad thing.
Boris the clearest sign yet that the Tory party should finally consider limiting their affirmative action admission policy giving undue preference to airheads and bimbos. Those people have valuable contributions to make in other areas, like bake-offs, internet video and, in this particular case, shampoo advertising. Sadly, the Tories' worthy experiment in inclusion has shown that those skills do not always readily translate well into public policy.
"Yeah well the gang from Redmond was also messing about with the usage-paradigm too but that doesn't seem to have bothered you too much. Win 8 and 10 (Redmond it seems can't count as they skipped 9) both have terrible User Interfaces."
There's no doubt the Metro experiment led to an upswing in the Linux curious. What they would have found is, well, 100 distros and about 50 shells to sort through, each with some annoyance or other. In the end it was easier to stay with Redmond, spend the 90 seconds installing Classic Shell, and carry on as if nothing had happened.
Interestingly, there are now far more instances of Ubuntu Server out there than Desktop. How may flavours of that are there? Two.
Unity was originally why I gave up on Linux as a desktop OS. Anyone over 25 and in regular employment, who therefore lacked the time to test or even read about 100 distros, and wanted a stable and sensible desktop with high adoption, just had to bail. That hasn't changed. Choice, bollocks. Look at this mess.
Cars are a lot faster than they used to be, and there are more of them about. Here is a list of 'data management' feature your car probably has. You've probably used some of them without even realizing it:
Adaptive power steering
Adaptive cruise control, adjustable, with various warnings
Electronic throttle (or more likely, fuel injection programming), for the
Automatic braking after a collision to prevent run-on collisions
Tyre pressure sensors
Lane detection with or without steering input
Blindspot object detection
Driver drowsiness detection
Intrusion detection & immobilization
Automatic hazard lights and emergency calls after collisions
If you do get a fast car without all these features, or you want to turn them off, then drive on the track, aka test/dev, where you can't hit anything. Track driving is a legitimate use case. It is also specialized and unusual.
Your OS has maybe two of these features. Your apps are drunk, libidinous teenagers who want to borrow the car. You figure it out.
Do you know what data you have?
Can you point to where your data are and who's reponsible for them?
Do you retire data?
Do you know what data you don't have?
Are your data available, reliable and compliant for the people who need it, assuming you know who they are?
Does your org understand the difference between privacy and confidentiality?
Does your management know what it means to pose an empirical question?
Is your management reasonably free from all of the following biases: survivor, confirmation, belief, recentness, egotism?
Does your org understand the difference between a marketing trend and an empirical finding?
Is your org capable of changing its mind in the face of emprical evidence?
Does your org tolerate short term failure and uncertainty?
If you answer no to any of these questions, it's best to learn to walk first.
Facebook is already full of sad people, no additional investment required:
This is a panel study, I don't think they tried to establish causation. But whether sad people flock to Facebook or Facebook makes them sad is, I think, a step further than I want to bother with.
This is explained by market dynamics. Acquisitions always attract a premium where there is a market (i.e. public information about the transaction, even if incomplete).
If HP offered market price, the publicly traded shares in the target would immediately jump to that price or higher, in anticipation of horse trading, competitive bids, and general betting by other actors. They would also attract other genuinely interested parties, who might make a competing offer and complicate and prolong the process. If the buyer offered 10% more than market price, the same would happen. The buyer has to deploy overwhelming force (40% premium here, iirc) to cut out speculators and other bidders and get the deal done quickly. It is worth paying for this.
The other dynamic is that the buyer has to use an investment bank to put the deal together, because there is operational knowledge required that the buyer doesn't have, and financial risk that must be underwritten, if only briefly. The bank is paid commission on the headline transaction price. They will inflate that price to the maximum plausible number, and although they will hesitate to intimidate, manipulate and short shrift their own customer to do it, that hesitation can be measured in milliseconds.
Vladimir Putin and Donald Trump.
It's all starting to make sense. The media deals, the risk appetite, the grandiose narrative of a love etched in the stars. They don't need a safe ship. They don't even need to go to the moon. They can make it true if their believers believe it enough.
If this is how Intel corporate customers get treated, spare a thought for us retail suckers who bought a 530 series SSD from Intel. The drive has a controller bug that Intel took three years to admit to. During that time, the replacement 535 came out with the same bug. Intel finally posted a patch. But its SSD toolbox does not apply the patch, it is not listed under 535 firmware updates, and is devilishly hard to google. The disclaimer on the patch states that it is applied at your own risk, so it's not clear what it does to your warranty.
Man who sleep with snake sleep long. I guess I have only myself to blame.
This is a very perceptive comment, if I may be so bold. These are people who can't think other than in fixed categories. You're either an immigrant or you're not. You're white or black, male or female, alive or dead (cf abortion). There's not much space in this world view for the epistemological humility and provisional knowledge that characterizes skepticism, the scientific method, open inquiry, or ambiguity. This may be why science in general, and evolution in particular, drives right wingers nuts.
This kind of digital boundary-based thinking 'maps' directly to geography and nationalism. The psychology of it is unfortunate, because these categories become bound up with identity, which humans use to hold their analogue brains together.
A useful aside is that thinking in simple categories both looks like common sense, and is very effective in management, because it is easy to package, prioritize, and communicate. Vague managers are bad managers. Categorizing things makes you look like strong and decisive. Loss-making products will be axed, Immigrants will be sent home, projects are either finished or they aren't, abortions will be stopped. Acknowledging that the world is complicated and all knowledge is provisional makes you look weak and indecisive, even if it's true.
We played this game when we moved to Wales in 2001. The 300 jobs we would create were all over the press. It was all meant well. Maybe 50 happened. No journos turned up in year three to check. Ford and GM are playing this game right now. Tomorrow it will be the tech industry.
Some factory moves will be announced and there will be pageantry. There will be some tax tweaks, and trade tweaks, that will be spun as a revolution for the working man (sic). In general, nothing will change, because it can't, except at the margin. This is not new.
I read some of the opinions on Fox News (do not try this at home). The 2016 election was about nationalism, not about economics. The forgotten middle the alt-right claim to speak for (whose jobs are supposed to have gone to Asia) is just collateral damage in a right wing identity crisis.
(Brexit is also a right wing identity crisis.)
The genius of Tr*** is that he doesn't have to do anything to claim victory, other than claim victory.
The article makes the useful point that a front-loaded ETL style education already looks like last year's hipster beard. This is where you extract a bounded corpus of data from trusted (old) sources, transform it to please your teacher, load it into your brain, and then dispense it at zero marginal cost and get paid forever. This may seem improbable in our sector, and it is partly a myth, but it is how large swathes of the population think of education - as something you can finish.
In contrast, I don't know whether learning how to learn is the answer to automation, but I don't have a better one. There are two problems.
The first is that although employers rely heavily on employees' meta-aptitude, that is, on their raw learning ability, they wouldn't recognize it if it bit them on the leg.
The second is that a generation of left behind people are at this moment being sold the notion by Donald Trump that if only we can make Apple make the iPhone at home, then ETL education will be good enough again, retrospectively. We'll just bring back the jobs like in pappy's day, and then productivity growth won't matter so much.
That is a flat lie. The truth is way scarier: If you don't have the neural infrastructure to bootstrap your way into Spanish or monotonicity or erosion or stakeholder management or royal prerogatives or Bayes or the Persian Empire or recursiveness from a standing start, no matter what you studied, you're a goner, Trump or no Trump.
P.S. You might be a goner even if you can.
Star Trek aside, perhaps we would save some person-years by managing a better characterization of enterprise IT. Enterprise IT is Boeing and Airbus, not XtremeAir, Lockheed Martin, or SpaceX.
In other words, the kicks are seldom where the money is. Flying an Airbus is mind-crushingly dull, and the tech is decades old. There's a reason for that.
You can go and work for a sexy company that builds cutting edge tech, but you won't sell many planes to Lufthansa. You decide.
A more IT-ish way to put this is that the NFRs in Enterprise are far harder to achieve than the FRs. This or that sexy tech is neither here nor there when the real differentiator is six sigma operations.
So what we are saying is that 3D X-Point, instead of being a pants-shaking revelation, might instead just be Quite Good.
Quite Good is nothing to sneeze at, when most things are, ipso facto, Fairly Average. If you can do nothing but reliably distinguish those two categories, you are of more use than all the marketing people and telephone sanitizers on the planet, collectively.
Rather than expound further, I would refer you to the Blackadder episode that exists to explain this phenomenon with uncommon precision. But I don't know which one it is.
Biting the hand that feeds IT © 1998–2019