Posts by Kevin McMurtrie 3560 publicly visible posts • joined 15 Jun 2007
I think there are a few indicators for imminent failure of a company:
- Abuses customers in favor of profit
- Focuses more on government politics than public opinion
- Apathy towards abuses of their services (or encourages it for profit)
- Shifts attention from architecting amazing products to architecting amazing offices
By these metrics, Google is following the path of imminent collapse that Yahoo blazed 20 years ago, while Yahoo could make a comeback for no better reason than sucking less than Google.
Seen it many times. Server get slower and slower, fails, recovers to full speed, then the cycle repeats. It's often caused by that one coder that isn't explicitly closing external resources - files, database connections, HTTP responses, etc. They run out and the server dies...but the error messages use just enough memory to invoke GC and it magically recovers. The knee-jerk reaction to add memory makes it much worse.
I'm all for creating an online universe, but this isn't it. Facebook is buggy as hell, slow, clumsy, and completely incapable of taking on next-generation functionality. The best it do is profit from abuse and chaos. This is like some crackpot with poor welding skills and no concern for physics claiming he's going to build a spaceship.
Too much spam on Usenet? You won't see it with the superior filtering on Google Groups. (99.99% of Usenet spam is from crime gangs using Google Groups)
Too much spam in your e-mail? You won't see it with the superior filtering on Google Gmail. (90% of email spam is from or returning to Gmail accounts)
Ads too slow? It's faster if Google hosts your content with AMP. (Google is jamming ad system outside of AMP)
It goes in waves. Apple has very long update cycles but they don't offer discounts on badly aging hardware. I really needed to update at a time when obsolete Trash Can Mac was still the premium offering. In came a couple of fast Linux desktops. They have a mix of spinning rust and SSD combined in ZFS giving me performance and storage I could never afford on a Mac with APFS.
They're talking about memory over network and Optane, so it's swap.
It sounds like the real product is:
- Fancy swap limited to containers so your software-defined-swap components don't need to be swap-safe themselves. That's freedom from a hefty limitation.
- A tool to gather up all your random bits of unused fast storage and pool it for container swap.
It seems like this product is only practical on systems with incredibly fast interconnects. Even so, the chance for complex patterns to appear in many simple components interacting with each other would keep me up at night. I'd hope for The Reg to test it for us but this makes me think they don't have that kind of equipment in-house.
Cloudflare and their fans like to use analogies like postal service, road builders, and power companies. Illegal stuff happens and it's nobody's fault. What if a road builder knows that they are building a road for a criminal to move stolen goods? What if the postal service knows that packages from a certain individual always contains stolen credit cards, illegal drugs, or weapons? How about if a customer asks for more power delivery to their illegal poppy greenhouse and opioid factory? Yeah, they'd all get in trouble.
Cloudflare has no defense because they know exactly who their criminal customers are and they willingly profit from it. Cloudflare even goes the extra mile to help their criminal customers stay anonymous after crimes have been reported to them.
At least once a month I get spams for clusters of obviously fake stores on Cloudflare. They have a cloned dozens of websites, JavaScript malware, display fake security certifications, and are taking credit cards with zero encryption. Cloudflare gives zero fucks when told about it. That's when they start breaking the law.
There must be some kind of stealthy active signal visualization for avoiding large objects.
Spread-spectrum radar? Wouldn't get much resolution at frequencies that will pass through sea water but might see big things that would hurt to run into.
Ultra-brief directional sonar pulses that can't be detected unless you know exactly when and where to listen for it?
How about scouting drones? Give them loud sonar and telemetry signals so they can be spread across a very large area. Everyone will hear the drones but have no idea where the listening sub is. These patrols are all for show so might as well break out the disco lights and mirror ball.
Leave the sub at home. Sharks with frickin lasers.
Generic solutions degrade as your requirements become more specific. AWS Serverless becomes exponentially complex when it comes time to scale-up the features. Even AWS hosting has a point where all of the complex costs and bewildering performance bottlenecks are more difficult to work within than running your own hardware.
Google will destroy themselves. They've set expectations of infinite hockey-stick growth but they've run out of things to plaster ads on and they've completely lost focus. Their "moon-shots" are failing and, other than some corporate cloud apps, their apps are in bad shape. Now they're desperate to squeeze the money out of everything. They want to cripple ad blockers in Chrome. I noticed that Android 11 has advertisements for Google Cloud storage and completely FUBAR microSD support.
Google built their wealth on open source software, ads, stolen personal information, and pirated media. As such, they own nothing but use by habit. The world can fork and move on.
There's very little software in the world worthy of subscription services, and auto manufacturers definitely aren't the ones producing it. VW's Car-Net subscription service is losing signal because the 3G cell bandwidth it needs is being recycled for LTE and 5G. People have only noticed because they're wondering if they'll get a free upgraded head unit. Nobody actually buys Car-Net.
V3 cripples it enough that people have already given up hope when V3.1 kills it.
Google has been plotting the same demise for many Android features that threaten advertising, spying, and cloud revenue. Android 11 is a dumpster fire and Android 12 is shaping up to be so bad that I wouldn't use it at all.
It must be a trick question. I can't imagine a revenue generating company settling for a 15% failure rate of anything. Three or four minor failures a year is enough to make new testing, review, and change control policies suddenly appear.
This would make for good patch cables, assuming fiber jacks ever become cheap enough to go mainstream.
Another would be scientific instruments needing extreme noise or voltage isolation.
I'm not so sure about automobile use. As others have said, the only advantage is reducing the need for EMP-safe signaling. That's not a big deal because every device still needs giant inductor-diode-resistor-capacitor power filters.
It's time for the big cellphone players to take Android away from Google. The OS started going to hell several versions ago. Each new revision has a slight UX improvement and massive new API restrictions. Right now, Google is hell-bent on breaking the filesystem so badly that only garbage game clones will be left in Play Store when they're done. If you have an microSD-friendly app with Android 11, you may have noticed that it's now 5x to 10000x slower because it's routed through Google's half-assed FUSE. Arbitrary file access will soon be revoked for any app not granted a magic exclusion by Google. In Android 12 there will be absolutely no non-Google way to back up or restore the phone. Even ADB warns that backup/restore is deprecated.
I haven't had a phone that could run LineageOS in a while but it ran a LOT better than anything from Google and the OEMs.
After working in a few programming languages and styles, I can confidently say that most asynchronous designs are antipatterns. There are few times when you call a method but don't need the answer until later. When most calls are synchronous by default, you're adding async wrappers 4% of the time. When they're async by default, you're adding unwrapping or chaining 96% of the time.
When I was a kid, the size of structures changed all the time. Maybe you're writing library version 3.0 but you know somebody's going to compile against 1.0, 2.0, 4.0, etc. where the size of things has changed. Fields get added. Integers change size. You had to pay attention to size fields or your code crashed.
Of course there was no Internet back then to provide instant updates. You typed in the hex code patches from a magazine because the planet was colder, bits were much heavier, and there were a million holders for disks on a desk but zero for carrying them through the snow of the constant California Bay Area blizzards. Nobody would even write a 161 MB word processor because, obviously, nobody was strong enough to carry that many bits in a backpack.
We can't work with just the good languages. Kotlin and Java observe features in trendy new languages then re-implement them in a way that actually works correctly. If you were shown side-by-side comparisons of Scala and Java, you might think that Scala is far more elegant. When you look at the language-level bugs in each, suddenly Scala is terrifying.
Will the Google Search app lose its spying capabilities if it's not used? How about the "Weather" apps that come with every phone to record GPS coordinates?
Google's has an ongoing war against things that don't make them money. Google made microSd performance 5 to 10000 times slower in Android 11. The logical next step is making sure that Android 11 forgets which apps you have disabled in ADB.
Quite the opposite. Lots of companies think Docker == free because they're using the open source parts. Pulls from Docker's image hosting is the only commercial product they're paying for now.
Docker's trick will be providing something that corporations WANT to buy, rather than charging for something that developers can replace. I can replace Mac Docker Desktop with an SSH tunnel, so that wasn't a good place for them to start.
Golang could be a valid problem in a highly interconnected codebase with a lot of complex data structures. Imagine all of those AWS fields and Amazon constantly making adjustments to them. Not even Amazon's own CloudFormation handles everything correctly. I do think Java 12, and especially 16, have a significant advantage there. Just stay away from the 3rd party frameworks that lead to complexity and dependency tech debt, like Lombok and Spring factories.
I still believe that the primary driving force behind eliminating the headphone jack was forcing phone owners to keep BT on and scanning tracking beacons. Since beacons only send a GUID, a the scans pass through a 3rd party service for conversion to coordinates.
There was a lot of marketing materials for these services about 20 years ago. Retailers could place BT beacons indoors to keep fine-grained navigation working without GPS signals. In return, retailers would know exactly where people were shopping by which beacon GUIDs were being looked up. The marketing materials have gone to a lower profile but the systems are in use. Turning BT scanning on makes Google Maps work in Japanese subways. A recent employer uses the beacons to keep office maps working without GPS.
Managing data in multiple places requires completely different application design. Each location has different performance and availability, and it's all dynamic.
AWS does recommend that everything be multi-region. Of course they would, because it's 3x money for them: 2x hosting + failover overhead + moving data
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
