« Designing for Resiliency will be so 2013 | Main | Ask HS: What will programming and architecture look like in 2020? »

Stuff The Internet Says On Scalability For December 28, 2012

It's HighScalability time:

  • 306 items per second: Orders on Amazon
  • Quotable Quotes:
    • @hackofalltrades: When positive change is only viewed through its scalability, bad things happen.
    • @faizanj: Is it time for #Netflix to move to a hybrid cloud architecture similar to Zynga zCloud?
    • @adrianco: we try not to invest in undifferentiated heavy lifting
    • @qui_oui: "scalability": a word that makes me think of how likely you are to have the ability to grow scales.
    • @Ninad_M: The question is, is #antifragile conceptually opposite of #bigdata
    • @pbailis: Batch your disk/network IO, kernel interrupts, customer package shipments -> delay arrival but increase efficiency
    • @Carnage4Life: One lesson that is hard for people to learn. Knowing that something occurred is different from knowing why it occurred
  • The best tech documentation both informs about the technology and teaches the wider context in which it plays a part. That fits the 400+ page Akka Documentation perfectly. In it you'll find excellent information on actors and the various architectures that can be created with them. Much to learn here. 
  • Lots of good full stack discussion on the different aspects of why page speed matters: CDN, DNS, static vs dynamic, geographical location, traffic spikes, DOM, rendering, network, app server, data layer vs presentation layer, problems with averages with Google Analytics, cached expiration, and more.
  • Great write up on the process used to make Node.js w/1M concurrent connections! Includes tweaks to /etc/sysctl.conf, increasing the heap limit and how to actually run a test that large.
  • Speaking of Node.js, here's Realtime Node.js App: A Stress Testing Story from Bocoup. A wonderfully detailed look at how to optimize a balance WebSockets with long polling. Lots of tricky subjects, including serialization, burst connection handling, and hidden connection limits.
  • Cloud-Powered GPS Chip Slashes Smartphone Power Consumption. Cool hybrid architecture where intelligence is applied to a bunch of cloud data to reduce power usage on mobile devices.
  • GitHub with an excellent description of their Downtime last Saturday. Software updates to aggregation switches caused network problems. The reversion process went bad and brought down links which caused heartbeats to fail which caused a race which made the file servers unavailable. Tricky stuff that requires some major system changes to fix. On Hacker News.
  • With Stanford's online programming class bigger is better. Students get a lot more help from other students and "A large class enables us to encourage the top students, and it gives them the opportunity to run with their own ideas."
  • Static websites are the new black now that S3 now has Root Domain Support. Pretty URLs, high reliability, high scalability, fast enough, and cheap. No more keeping a webserver around just to serve static content. Now you just need an API for any dynamic portions. Also, How I Made my Blog 2.3x Faster
  • The perfect IRS tax storm: Tax season delay could mean scalability questions for the IRS...IRS may delay the start of filing, but they cannot extend the 4/15 due date. That could mean a flood of returns in a short time hitting their servers. They recently switch from batch to real-time processing.
  • Can we learn anything from how the body heals? Physics of Wound Healing I: Energy Considerations: Wound healing is a complex process with many components and interrelated processes on a microscopic level. This paper addresses a macroscopic view on wound healing based on an energy conservation argument coupled with a general scaling of the metabolic rate with body mass.
  • Why Access Patterns Matter in GPU Programming: it is important to structure your application so that different threads do not access the same bank of memory at the same time. In other words, you need to avoid bank conflicts.
  • MessagePack: looks like an interesting way to make JSON efficient. #serializationisevil
  • Amazon has  a cli: awscli, providing a unified command line interface to Amazon Web Services. It covers a good chunk of the services and looks like heaven for scripting.
  • Quora: Why did twitter move away from Ruby on Rails? That you get better travel miles with Scala was quite a surprise.
  • Cloud Computing:  U.S. Tax Compliance  for Complexity for Foreign Subsidiaries. The outdate tax law can't keep up with tech advances, leaving companies with uncertainty, especially on cross border deals. What a surprise.
  • Still confusing even when well explained...It's "locking" if it's blocking: "locking" and "lock-free" are two equivalent ways to order events, the difference being that with locking, you have to deal with suspension, while with lock-free updates, you have to deal with the effects of concurrent interferences.
  • Big Data Problems in Monitoring at eBay. Hit this post for detailed summary the talk: Entropy is important to look at; remove it as early as possible; Data distribution needs to be flexible and elastic; Storage should be optimized for access patterns.
  • Erlang on Xen: less than 100 milliseconds after the launch a new node starts to run the application code; relies on external elastic resource management services instead of services of a local operating system; No instances are running waiting for requests.
  • Good to see: Windows Azure Gets More Competitive With AWS, Launches New Mobile Features, Website Scaling, Data Sync And More

Reader Comments

There are no comments for this journal entry. To create a new comment, use the form below.

PostPost a New Comment

Enter your information below to add a new comment.
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>