dotScale 2013
who it's for
platform engineers, DevOps engineers, site reliability engineers, cloud engineers, backend engineers, infrastructure architects, database administrators
topics
MySQL replication and sharding; Elasticsearch cluster architecture; Docker container workflows; OpenStack cloud platform; Hadoop batch processing; Consul service discovery; TRILL-based virtual networking; OpenZFS storage management; cloud messaging systems; Google Cloud Platform experiments; dynamic cloud autoscaling; healthcare.gov operational lessons
| title | author | length |
|---|---|---|
| Scaling MySQL at Wordpress.com | Barry Abrahamson | 21m |
wordpress.com runs about 50 million sites backed by roughly 500 million tables across 120 shards and ~2,200 servers (≈500 DB servers), serving hundreds of… | ||
| Scaling, what you need to know | Quentin Adam | 18m |
Advocates splitting application into process and storage with concrete choices: PostgreSQL for transactional rows, Redis for TTL sessions, S3 (or similar) for… | ||
| How Draw Something absorbed 50 million new users in 50 days | Tugdual Grall | 05m |
The claim is concrete: 50 million new users in 50 days. The concrete tech named is Couchbase and its document-oriented NoSQL model. Expect specifics about the… | ||
| Zettascale | Ori Pekelman | 05m |
Claim: zettascale means zettabyte = 10^21 and current clusters of raspberries and commodity drives scale to terabytes and petabytes but not to zetabytes… | ||
| Uses of tmux explained | Stanislas Polu | 04m |
tmux runs as a client-server via Unix domain socket and creates pseudo-terminals with openpty so sessions persist when the client disconnects; detach and… | ||
| Welcome to your new job | Nicolas Fonrose | 04m |
Claims that every cloud action has direct cost impact: starting or forgetting servers, allocating disks, and architectural choices change bills. Recommends… | ||
| Why we built ElasticSearch | Shay Banon | 16m |
ElasticSearch exists to let users ask ad hoc questions over mixed data and get answers in milliseconds by collocating search, filtering and aggregation. It… | ||
| Introduction to Hadoop | Doug Cutting | 24m |
History and claim: Hadoop grows out of Nutch and Google papers and becomes a working platform that moves from 20 to 40 machines to thousands at Yahoo, handling… | ||
| The origins of OpenStack | Joshua McKenty | 25m |
History of OpenStack explains how a NASA/Rackspace project called Pan/Nova becomes a million‑line, 10,000‑person codebase by shipping fast not standardizing… | ||
| DevOps at scale | Jonathan Weiss | 15m |
Advice: plan for failure and automate the responses. Use AWS OpsWorks and EC2 APIs to measure key metrics and gate rollouts via onebox deployment before wider… | ||
| Evolution of service deployment | Dave Neary | 04m |
Chronicles a concrete shift: 1990 procurement cycles that take a year with IBM RS/6000 running AIX 3 with 100MB disk give way to 1996 developer-built servers… | ||
| Virtual Network over TRILL | Thomas Stocking | 19m |
Gandi.net U.S. R&D builds a production system that uses TRILL plus a VNI tag to turn VLAN limits into 16 million virtual networks and to hide VM MACs inside… | ||
| Software I'm excited about | Brad Fitzpatrick | 21m |
Names concrete tech and outcomes and whether they work Go is pitched as a pragmatic compiled language with lightweight goroutines and static binaries that let… | ||
| Why we built Docker | Solomon Hykes | 21m |
Docker presents a concrete shipping format and toolchain: a portable container engine that standardizes images, builds and runs them, and is already used at… | ||
| User Space in Cloud Platforms | Noah Zoschke | 18m |
Heroku explains how apps run down the stack from MRI Ruby through LXC containers on Ubuntu into cgroups and namespaces and then into Amazon VMs using the… | ||
| Experimenting with Things using Google Cloud Platform | Mandy Waite | 19m |
Deploying 500 sensor Moes (Arduino Leonardo R3 with custom shields: temp/pressure, audio, humidity, light; 300 had specialty sensors — 100 RF, 100 motion via… | ||
| Everything I learned about dynamic scaling of cloud apps. | Laurence Moroney | 19m |
Built a second‑screen quiz for the One Direction seven-hour livestream that faces million‑fan spikes every ten minutes, running on Google App Engine and using… | ||
| OpenZFS | Matthew Ahrens | 16m |
Claims enterprise-grade correctness by design: checksums all data and metadata and uses copy-on-write on-disk format so fsck is unnecessary. Implements RAID‑Z… | ||
| My Take on PHP | Fabien Potencier | 18m |
Demonstrates concrete work: Composer, HHVM, ReactPHP and Symfony components drive measurable gains; a PHP-based server on nginx reports 15× speedup for fat… | ||
| Building Robust Systems With Consul | Mitchell Hashimoto | 17m |
Consul provides DNS service discovery plus an HTTP API that returns only healthy IPs, a replicated KV-backed dynamic config store that agents cache and can… | ||
| Little fun with cloud API comparison | Laurent Cerveau | 05m |
Comparison lists concrete API call counts and behaviors across providers: Amazon AWS requires about 42 requests for a 40–50–element infrastructure because… | ||
| Iris: Simplifying cloud messaging | Péter Szilágyi | 06m |
Iris earns attention only if it states a concrete claim about cloud messaging and backs it with evidence: throughput numbers and latency under load, the actual… | ||
| Today we make the Internet of tomorrow | Michiel Leenaars | 07m |
A private foundation gives microgrants to concrete projects and claims measurable wins: funding for GPL v3, DNSSEC work, XMPP/IETF/W3C standards, Tor GNUnet… | ||
| boot2docker | Steeve Morin | 05m |
A tiny Linux that runs Docker as its only userland: a 25 megabyte ISO based on Tiny Core Linux that boots in three seconds and runs entirely from RAM so… | ||
| Security and Risks | Alison Gianotto | 22m |
Shows that breaches spike: 8 mega breaches in 2013 vs 1 in 2012 (700% increase) and daily attacks rise from 190k (2011) to 570k (2013); cites Adobe (152M) and… | ||
| Extensibility as a Product | Jean-Michel Lemieux | 19m |
Claims that prototyping shrinks setup from pages of docs into minutes and that teams scale by shipping extensibility as a product are backed by specifics: a… | ||
| Darwin & Protocols | Paul Mockapetris | 28m |
History shows DNS wins by simplicity but now reveals sharp limits: cache poisoning appears within a decade and Dan Kaminsky’s statistical attack forces stopgap… | ||
| Why did I create Salt? | Thomas Hatch | 19m |
Salt exists to manage entire stacks at scale and actually does it: benchmarks dramatically faster and proven across tens of thousands of servers. It combines a… | ||
| Building the Ideal Developer Platform | Jeff Lindsay | 19m |
Open-source platform called Flynn claims a developer release and boots a container compute grid by running Docker on hosts, then bootstraps services and… | ||
| Life in the Trenches of healthcare.gov | Robert Kennedy | 27m |
Deployment stories with timestamps and root causes not abstracts: permission changes at 10:48 a.m. revoke the app user and cause 100% errors while latency… | ||
| The Wikimedia infrastructure | Faidon Liambotis | 16m |
Wikipedia serves about 500 million unique users and 20 billion monthly page views peak at ~190,000 HTTP requests per second and runs ~1,000 servers across two… | ||
| nothing matches | ||