addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-checkcircle-with-crosscircle-with-pluscrossdots-three-verticaleditemptyheartexporteye-with-lineeyefacebookfolderfullheartglobegmailgooglegroupsimageimagesinstagramlinklocation-pinm-swarmSearchmailmessagesminusmoremuplabelShape 3 + Rectangle 1outlookpersonJoin Group on CardStartprice-ribbonImported LayersImported LayersImported Layersshieldstartickettrashtriangle-downtriangle-uptwitteruseryahoo

Cassandra at Rally Software

  • May 1, 2013 · 6:30 PM

Hello Colorado Cassandra Users,

Gnip has kindly offered to host our next meetup on May 1st, 2013.

We will be enjoying an engaging presentation by Steve Neely, Research and Development Engineer at Rally Software!

Be sure to come straight from the office, as food & beverage will be provided.


After a successful experience employing Cassandra for session replication, we chose to use it for backing our new authentication service. But we don’t have big data. Crazy? Maybe...

In this presentation we’ll explain why we chose Cassandra to store our relatively small data set and what Cassandra gave us --- HA, happiness and headaches. We’ll describe our service requirements, technology stack, data model, optimizations, the pains, and successes from our experiences using Cassandra.

Hopefully our story can help guide you in future projects using Cassandra data stores!


_____Tentative Outline_____

  1. Introduction

  2. Cassandra for session replication

    1. how it works

    2. write heavy

  3. Zuul authentication service and its requirements

    1. response times

    2. high availability

    3. multi-datacenter

  4. Technology choices

    1. JVM -- Java / Clojure

    2. netty

    3. Hector -- looked at Astyanax in its early days

  5. Data model

    1. denormalizing users-by-username

  6. Zuul is read heavy -- auth keys

  7. Optimization -- reverse column families

  8. Optimization -- caches

    1. row cache did not help

    2. implemented our own

  9. High availability

    1. game days

    2. shutting down production nodes showed we were misconfigured

    3. switching data centers success

    4. show request graph

  10. Debugging

    1. cassandra-cli and cqlsh

    2. datastax enterprise

    3. datastax support

    4. splunk, ganglia, gdash

  11. Future

    1. restructuring data model -- more denormalizing, use values as keys + empty value holders

    2. SLA requirements changing with oauth



Join or login to comment.

  • Jeff S.

    Not going to tempt the weather gods this evening...

    May 1, 2013

  • Joe M.

    Sorry, have to cancel for weather/traffic related reasons.

    May 1, 2013

36 went

Our Sponsors

  • DataStax

    Providing Speakers, Venues, Food, Beverages and Swag

  • Planet Cassandra

    Food, Beverage, SWAG, Speakers and Content

  • Gnip

    Speakers, venue, food, & beverage

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy