Hey, it's HighScalability time:

 

• 2+ billion: Google lines of code distributed over 9+ million source files; $3.6 bn: lower Google taxes using Dutch Sandwich; $14.6 billion: aggregate value of all cryptocurrencies; 2x: graphene-fed silkworms produce silk that conducts electricity; < 100: scientists looking for extraterrestrial life; 48: core Qualcomm server SoC; 455: original TV series in 2016;


• Quotable Quotes:
  • Ben Thompson~ It's so easy to think of tech with an 80s mindset with all the upstarts. We still glorify people in garages. The garage is gone...Our position in the world is not the scrappy upstart. It is the establishment.
  • The Attention Merchants: True brand advertising is therefore an effort not so much to persuade as to convert. At its most successful, it creates a product cult, whose loyalists cannot be influenced by mere information
  • @seldo: Speed of development always wins. Performance problems will (eventually) get engineered away. This is nearly always how technology changes.
  • @evgenymorozov: How Silicon Valley can support basic income: give everyone a bot farm so that we can make advertising $ from fake traffic to their platforms
  • @avdi: Apple has 33 Github repos and 56 contributors. Microsoft now has ~1,200 repos and 2,893 contributors.
  • Peter Norvig: Understanding the brain is a fascinating problem but I think it’s important to keep it separate from the goal of AI which is solving problems ... If you conflate the two it’s like aiming at two mountain peaks at the same time—you usually end up in the valley between them .... We don’t need to duplicate humans ... We want humans and machines to partner and do something that they cannot do on their own.
  • Brave New Greek: Unbounded anything—whether its queues, message sizes, queries, or traffic—is a resilience engineering anti-pattern. Without explicit limits, things fail in unexpected and unpredictable ways. Remember, the limits exist, they’re just hidden. By making them explicit, we restrict the failure domain giving us more predictability, longer mean time between failures, and shorter mean time to recovery at the cost of more upfront work or slightly more complexity.
  • Naren Shankar (Expanse): Everybody feels like they can look at the show and find parts of themselves in it. When you can give people collective ownership of the creative product you get the best from people. At the end of the day it shows. People work their asses off and accomplish the impossible.
  • Richard Jones: a corollary of Moore’s law (sometimes called Rock’s Law). This states that the capital cost of new generations of semiconductor fabs is also growing exponentially
  • Waterloo: His [Napoleon] strategy was simple. It was to divide his enemies, then pin one down while the other was attacked hard and, like a boxing match, the harder he punched the quicker the result. Then, once one enemy was destroyed, he would turn on the next. The best defense for Napoleon in 1815 was attack, and the obvious enemy to attack was the closest.
  • Daniel Lemire: beyond a certain point, reducing the possibility of a fault becomes tremendously complicated and expensive… and it becomes far more economical to minimize the harm due to expected faults
  • @greglinden: “For some products at Baidu, the main purpose is to acquire data from users, not revenue.” — @stuhlmueller
  • strebler:  Deep Learning has made some very fundamental advances, but that doesn't mean it's going to make money just as magically!
  • sulam: Twitter clearly doesn't have growth magic (or they'd be growing faster) -- but is that an engineer's fault? At the end of the day, any user facing engineering is beholden to the product team. Engineers at Twitter can run experiments, but they can't get those experiments shipped unless a PM is behind it.
  • Gil Tene: The right way to read "99%'ile latency of a" is "1 or a 100 of occurrences of 'a' took longer than this. And we have no idea how long". That is the only information captured by that metric. It can be used to roughly deduce "what is the likelihood that 'a' will take longer than that?". But deducing other stuff from it usually simply doesn't work.
  • @esh: Unheralded tiny features like AWS Lambda inside Kinesis Firehose streams replace infrastructure monstrosities with a few lines of code
  • @postwait: Listening to this twitter caching talk... *so* glad my OS doesn't even contemplate OOMs. How is that shit still in Linux? A literal WTF.
  • SomeStupidPoint: Mostly, it was just a choice to save $1-2k on a laptop (every 1-2 years) and spend the money on cellphone data and lattes.
  • @timbray: Oracle trying to monetize Java... Golang/Rust/Elixir all looking better. Assume all JVM langs are potential targets.
  • Kathryn S. McKinley: In programming languages research, the most revolutionary change on the horizon is probabilistic programming, in which developers produce models that estimate the real world and explicitly reason about uncertainty in data and computations. 
  • cindy sridharan: Four Golden Signals 1) Latency 2) Traffic 3) Errors 4) Saturation
  • @FioraAeterna: as a tech company grows in size, the probability of it developing its own in-house bug tracking system approaches 1
  • The Attention Merchants: In 1928, Paley made a bold offer to the nation’s many independent radio stations. The CBS network would provide any of them all of its sustaining content for free—on the sole condition that they agree to carry the sponsored content as well


• philips: Essentially I see the world broken down into four potential application types: 1) Stateless applications: trivial to scale at a click of a button with no coordination. These can take advantage of Kubernetes deployments directly and work great behind Kubernetes Services or Ingress Services. 2) Stateful applications: postgres, mysql, etc which generally exist as single processes and persist to disks. These systems generally should be pinned to a single machine and use a single Kubernetes persistent disk. These systems can be served by static configuration of pods, persistent disks, etc or utilize StatefulSets. 3) Static distributed applications: zookeeper, cassandra, etc which are hard to reconfigure at runtime but do replicate data around for data safety. These systems have configuration files that are hard to update consistently and are well-served by StatefulSets. 4) Clustered applications: etcd, redis, prometheus, vitess, rethinkdb, etc are built for dynamic reconfiguration and modern infrastructure where things are often changing. They have APIs to reconfigure members in the cluster and just need glue to be operated natively seemlessly on Kubernetes, and thus the Kubernetes Operator concept


• Top 5 uses for Redis: content caching; user session store; job & queue management; high speed transactions; notifications.


• Is machine learning being used in the wild? The answer appears to be yes. Ask HN: Where is AI/ML actually adding value at your company? Many uses you might expect and some unexpected: predicting if a part scanned with an acoustic microscope has internal defects; find duplicate entries in a large, unclean data set; product recommendations; course recommendations; topic detection; pattern clustering; understand the 3D spaces scanned by customers; dynamic selection of throttle threshold; EEG interpretation; predict which end users are likely to churn for our customers; automatic data extraction from web pages; model complex interactions in electrical grids in order to make decisions that improve grid efficiency;sentiment classification; detecting fraud; credit risk modeling; Spend prediction; Loss prediction; Fraud and AML detection; Intrusion detection; Email routing; Bandit testing; Optimizing planning/ task scheduling; Customer segmentation; Face- and document detection; Search/analytics; Chat bots; Topic analysis; Churn detection; phenotype adjudication in electronic health records; asset replacement modeling; lead scoring;  semantic segmentation to identify objects in the users environment to build better recommendation systems and to identify planes (floor, wall, ceiling) to give us better localization of the camera pose for height estimates; classify bittorrent filenames into media classify bittorrent filenames into media categories; predict how effective a given CRISPR target site will be; check volume, average ticket $, credit score and things of that nature to determine the quality and lifetime of a new merchant account; anomaly detection; identify available space in kit from images; optimize email marketing campaigns; investigate & correlate events, initially for security logs; moderate comments; building models of human behavior to provide interactive intelligent agents with a conversational interface; automatically grading kids' essays; Predict probability of car accidents based on the sensors of your smartphone; predict how long JIRA tickets are going to take to resolve; voice keyword recognition; produce digital documents in legal proceedings; PCB autorouting.

Don't miss all that the Internet has to say on Scalability, click below and become eventually consistent with all scalability knowledge (which means this post has many more items to read so please keep on reading)...