Dan HarveySystems Architect
Revolutionising the world of research
with Amazon Web Services
Mendeley helps researchers work smarter
Mendeley makes science morecollaborative and transparent
Sync between computersand peers around the world
1 million researchers...
Uploaded 130 million citations
40 million unique
17 million papers
16 TB of data
Document Storage
• Backed on to S3
• Previews for 16TB of PDFs?
PDF Previews
ElasticBeanstalk S3
Process Queue
Load PDFs
Store PNGServe
via Cloud Front
Render to PDF
System Overview
EMREC2
S3
EBS
RDS
VPC
Article Search
• Based on Solr (open source search)• 40GB index• Variable usage
Day of the month
H-pRequests
Solr Layout
SolrSlave Solr
SlaveSolr
Slave
EBS
ElasticLoad Balancer
SolrMaster
SearchQueries
Inside VPC
Outside VPC
• Machine learning on Hadoop
• Personalised article recommendations• Collaborative filtering based
• Running on Elastic Map Reduce
Summary
• Not all or nothing
• Focus on your problemnot “Undifferentiated heavy lifting”
- Werner Vogels
• Learn the building blocks AWS provide
Enjoy what you’ve seen?
We’re hiring!
Senior Java Engineers
chat to me afteror e-mail/tweet