Stuff The Internet Says On Scalability For March 4, 2011

Submitted for your reading pleasure on this beautifully blue and sunny Friday...

  • @Werner: Each day #AWS adds enough computing muscle to power one whole Amazon.com circa 2000, when it was a $2.8 billion business http://wv.ly/gMr8LQ
  • Building servers to rule in hell. Datacenters spend a lot of energy on cooling down processors. Why can't they operate at higher temperatures? This is the proposition addressed by James Hamilton in Exploring the Limits of Datacenter Temprature and Datacenter Knowledge in What’s Next? Hotter Servers with ‘Gas Pedals’.
  • Quotable Quotes for 200 Watson:
    • @jreichhold: One thing working at Twitter teaches me daily is that all scale is relative. What seemed impossible last year is now the daily case.
    • @dannycast0nguay: If you’re concerned about scalability, any algorithm that forces you to run agreement will eventually become your bottleneck.—Werner Vogels
    • @rael: No shortcut ever goes undetected by scale.
    • @sramji: Caching is for scalability, not for speed. Your cache *will* break and if your backend is not fast you are sunk. - Dan Hobbs #apimeetup
    • @joestump: OH: "I love the smell of data redundancy with automatic failover in the morning."
  • Scalability Porn:
  • Jeremy Zawodny with an detailed explanation of how Craigslist implements sharding on Redis using 10 hosts and 120GB of RAM.
  • Volley: Automated Data Placement for Geo-Distributed Cloud Services. Murat Demirbas with great coverage on a paper by Microsoft Research on a heuristic strategy for data placement to geo-distributed datacenters. The evaluations show that Volley converged after a small number of iterations, and reduced skew by 2x, inter-datacenter traffic by 1.8x, and latency by 30%. 
  • Death match! EBS versus SSD price, performance, and QoS. Baron Schwartz with ring side commentary finding in a knock-down-drag-out fight between the EBS gang and the SSD thugs, a small number of SSDs mops the floor with the competition, and walks away with the prize.
  • If like nearly everyone else you are confused about security, Everything you need to know about cryptography in 1 hour, just may help you become slightly less confused, but probably not. 
  • In a world where algorithms are for people who don't know how to buy RAM, isn't 24 Gigabytes of Memory Ought to be Enough for Anybody? Jeff Atwood observes: To me, it's more about no longer needing to think about memory as a scarce resource, something you allocate carefully and manage with great care. There's just .. lots.
  • Scott Lystig Fritchie with a nicely detailed report on Webmail for Millions, Powered by Erlang: The architecture and lessons learned implementing a webmail system in Erlang, using UBF and Hibari, a distributed key-value store, to accommodate a large user base.
  • On the Engineering of SaaS. Theo Schlossnagle shares insight into what’s in store for you as you transition from a software company into a SaaS company: A customer of one;  You aren’t a software company anymore; Continuous Deployment; Quality Assurance is now a continuous process; Multi-tenancy design.
  • Ruby on Rails: Scaling Your App for Rapid Growth. Jolie O'Dell interviews Gowalla CTO and co-founder Scott Raymond on their experiences with RoR. So, as a developer for a popular service, the challenge becomes trying to foresee which actions will be the most frequently requested, which data types will be the fastest growing, and which actions are the most performance-sensitive.
  • Don't look now, but Google has replaced your HTTP with SPDY. Don't worry, it's good for you. SPDY is an experiment with protocols for the web.  Its goal is to reduce the latency of web pages.
  • Packet Pushers on Breaking the Three Layer ModelIs this the point we start to re-think the 3 Layers of Network Architecture, will the future generation laugh at core, distribution and access designs? Yes.
  • A list of Our favorite Cloud, SaaS, and Tech bloggers! you may be interested in following.
  • Curt Monash with a nice update on some new DbShards features: dbShards update and Clarification on dbShards’ shard replication. Plus a good overview on what the Membase CouchOne merger really means. 
  • Baron Schwartz runs some benchmarks to test Is VoltDB really as scalable as they claim? The answer is yes: I’ll tell you: scaling a synchronously replicated, active-active master, fully ACID, always-consistent database to a 40-server cluster is impressive