Abstract | ||
---|---|---|
With the advent of Twitter, the follow model has become pervasive across social networks. The follow model enables users to follow other users i.e. subscribe to content created by other users, thereby, establishing the concept of a following feed for a user. At Pinterest, we continually store, update and serve feeds for millions of users and fan out millions of newly created pins/repins to thousands of followers, leading to billions of operations everyday. We describe the current feed storage solution, backed by Apache HBase, at Pinterest. We describe how we handle data management challenges unique to our scale, in the wake of strict performance and availability requirements. We also present a qualitative comparison to our previous "following feed" architecture, backed by Redis. |
Year | DOI | Venue |
---|---|---|
2013 | 10.1109/SocialCom.2013.7 | BigData Conference |
Keywords | DocType | ISSN |
social network,apache hbase,qualitative comparison,scaling deep social feeds,current feed storage solution,following feed,strict performance,availability requirement,data management,sql,content management | Conference | 2639-1589 |
Citations | PageRank | References |
0 | 0.34 | 7 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Varun Sharma | 1 | 462 | 21.91 |
John M. Carroll | 2 | 4950 | 1233.96 |
Abhi Khune | 3 | 0 | 0.34 |