Twitter replicates twice in data centre move

 

Two-stage migration.

Twitter has completed a major migration of its infrastructure into a new data centre, shifting 20TB of tweets and much of its backend gear twice in the process.

The company's engineering team said the migration process began last September to increase resilience and reduce downtime for the microblogging service.

Engineers started by extending Twitter's systems to replicate tweets to multiple data centres – allowing the service to operate throughout the migration.

Some hardware and "non-runtime systems" were shifted from the old primary facility to a secondary data centre, which also acted as a "staging laboratory" to test Twitter's replication and migration strategy.

Engineers started serving live traffic from the secondary data centre for test purposes "and to continue to shed load from our primary data centre".

At the same time, engineers prepared a third location which was to become Twitter's new home.

Once the migration strategy from primary to secondary was proven, engineers shifted "all of Twitter from the first and second data centres to the final nesting grounds" - the new third data centre.

"This essentially required us to move much of Twitter two times," engineers said.

Copyright © iTnews.com.au . All rights reserved.


Twitter replicates twice in data centre move
 
 
 
Top Stories
The True Cost of BYOD - 2014 survey
Twelve months on from our first study, is BYOD a better proposition?
 
Photos: Unboxing the Magnus supercomputer
Pawsey's biggest beast slots into place.
 
ANZ looks to life beyond the transaction
If digital disruptors think an online payments startup could rock the big four, they’ve missed the point of why people use banks, says Patrick Maes.
 
 
Sign up to receive iTnews email bulletins
   FOLLOW US...
Latest Comments
Polls
What is delaying adoption of public cloud in your organisation?







   |   View results
Lock-in concerns
  29%
 
Application integration concerns
  3%
 
Security and compliance concerns
  28%
 
Unreliable network infrastructure
  9%
 
Data sovereignty concerns
  22%
 
Lack of stakeholder support
  3%
 
Protecting on-premise IT jobs
  4%
 
Difficulty transitioning CapEx budget into OpEx
  3%
TOTAL VOTES: 1105

Vote