Introduce myself and what I do at OWW ( Who is OWW and who are our customers)
What are we storing in CouchbaseSide note ( this doubled after adopting CB)
What did we require when looking for a single caching / NoSQL solution
What works and what does not for OWW standardization efforts as we assumed without knowing our needs. How picking the right HW/SW combinations along with tweaking make all the difference.
From pre-prod to prod; why what you test on matters in production releases
Explores communicating with developers and understanding what they require from Couchbase as well as making sure the developers know all the capabilities available.
Examines what happens when systems are not properly configured for Couchbase operational requirements.
Which combinations of couchbase bucket and memcached buckets work well in a cluster.What can your hardware handleWhere is NoSQL appropriateHow to balance # of clusters with available admin resources ( alerting / failover etc)
How best to create a Highly available cache solution without wasting resources.Summarize lessons OWW learned.
Describing some of the applications in testing
The key to a successful implementation of Couchbase, is to begin from the ground up. Know the capabilities of the hardware you choose to work with as well as the operating system performance and benchmarks. Disk, memory, network and base operations, all require resources that will effect the performance of your cache. Know the limitations you face before over committing your system to caches it is unable to properly handle without proper configuration and design.
Transcript of "Orbitz_Couchbase_SF_2013"
From Sectional to Sofa
NoSQL with .
J. Steven Young
Show Me the Cache
• 11 Clusters (4 mirrored)
– 100 nodes
• > 3 TB of data
• ~430m objects (146m in largest)
• Total ops/sec ~ 75k
– *149k with HA
In The Beginning
• Cache Scalability
• Drop in replace Memcached
• Database relief
• HA caching
From the Bottom Up