Hey, it's HighScalability time:

What a beautiful example of Moore's law visualized through the evolution of Lara Croft! (from @silenok)

• $1 million: per day gross of Clash of Clans
• Quotable Quotes:
  • @dancow: In 45 minutes, the largest trader in U.S. equities went bankrupt because of bad devops
  • @bmdhacks: How to be a 10x engineer: Incur technical debt fast enough to appear 10x as productive as the ten engineers tasked with cleaning it up.
  • @CompSciFact: Scaling poorly: Performance degrades with problem size
    Poorly scaled: Things change far more rapidly in one direction than others
  • @mikiobraun: Before scaling out, a machine learning person would always try some approximation shortcut to achieve speed up. #cheating #orisit
  • @cshirky: 3/4 If your organization has ever made a significant and unpleasant change based on something you measured, you can probably use more data.
  • @PatrickMcFadin: Service Discovery Overview: ZooKeeper vs. Consul vs. Etcd vs. Eureka 
  • @jaykreps: TIL: Dequeuing a single item in RabbitMQ requires traversing every single item in the queue. Oh my.
  • @Carnage4Life: No single recipe 4 success. Great companies had bad habits; Apple micromanagement, Google random side projects & Facebook used fricking PHP
  • Stubbornly Persistent: although life would persist in the absence of microbes, both the quantity and quality of life would be reduced drastically.


• At inflection points change the world must. Netflix: In the early days of Netflix streaming, circa 2008, we manually tracked hundreds of metrics, relying on humans to detect problems.  Our approach worked for tens of servers and thousands of devices, but not for the thousands of servers and millions of devices that were in our future.  Complexity and human-reliant approaches don’t scale; simplicity and algorithm-driven approaches do.


• IBM is turning Watson into a platform, offering 5 new services: Speech to Text, Text to Speech, Visual Recognition, Concept Insights, Tradeoff Analytics. GA probable next month. Good discussion on Hacker News. Most of the services allow for training through feedback. Some question the quality of the services, but it's early days. Pricing is not set. Hopefully it won't suffer from what these next gen deep learning services tend to suffer from: expensivitis. Who can afford $1.00 per 1000 API calls for a mobile app that needs to acquire users? IBM, make it cheap, try for ubiquity. Cool stuff will happen.


• Looking for that next step in distributed reliability? Look at TLA+. Murat has several articles on TLA+ and is using it his teaching distributed systems class. Oh, TLA stands for Temporal Logic of Actions. Leslie Lamport has many papers on TLA. James Hamilton wrote up their experiences at Amazon using TLA+: Challenges in Designing at Scale: Formal Methods in Building Robust Distributed Systems: TLA+, a formal specification language invented by ACM Turing award winner, Leslie Lamport. TLA+ is based on simple discrete math, basic set theory and predicates with which all engineers are quite familiar. A TLA+ specification simply describes the set of all possible legal behaviors (execution traces) of a system. 

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)...