Polyglot Paralellism: A Case Study in Using Erlang and Ruby at Rackspace
summary
Rackspace rewrites a failing Ruby/Perl stack into a polyglot system that backs up 20,000 devices across three continents and eight datacenters with <1% failure rate daily. Rails provides the front end, MongoDB stores blob backups, and an Erlang backend runs a runner/worker batch framework where monitored workers auto‑restart on crash. Webmachine exposes a REST API. The concrete wins are throughput from massive parallel SSH, smaller DB queries, and seamless recovery when VMs fail.
AI-written summary · about
chapters (31)
- 0:00Introduction
- 0:50Problem Part 1
- 1:26Devices
- 3:37Blobs
- 8:21The Solution
- 14:26Application Architecture
- 15:15History of Erlang
- 15:54HD301 ATM Switch
- 16:11Uptime
- 16:52Erlang
- 18:07Single Assignment
- 19:49Immutability
- 21:06Dictionary
- 22:08concurrency oriented
- 23:16quicksort
- 25:37job framework
- 25:56runner and worker communication
- 28:02behavior
- 30:07spawn
- 31:21worker process
- 31:56story time
- 33:06RESTful API
- 34:12HTTP Web Machine
- 35:39Dispatch Rule
- 36:25Resource Module
- 37:03Resource Functions
- 38:03Requests
- 38:33Content Types
- 39:51Primitive of Set Obsession
- 40:14String IO Lists
- 42:04Outro
