Name
Abstract | ||
|---|---|---|
Migrating large-scale data sets (e.g. social graphs) from cluster to cluster and meanwhile providing high system uptime is a challenge task. It requires fast bulk import speed. We address this problem by introducing our "Demand-driven Bulk Loading" scheme based on the data/query distributions tracked from Facebook's social graphs. A client-side coordinator and a hybrid store which consists of both MySQL and HBase engines work together to deliver fast availability to small, "hot" data in MySQL and incremental availability to massive, "cold" data in HBase on demand. The experimental results show that our approach enables the fastest system's starting time while guaranteeing high query throughputs. |
Year | Venue | Keywords |
|---|---|---|
2014 | ADVANCES IN DATABASES AND INFORMATION SYSTEMS (ADBIS 2014) | Bulk loading,HBase,MySQL |
Field | DocType | Volume |
Graph,Data set,On demand,Computer science,Demand driven,Database,Distributed computing | Conference | 8716 |
ISSN | Citations | PageRank |
0302-9743 | 1 | 0.38 |
References | Authors | |
10 | 2 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Weiping Qu | 1 | 6 | 3.15 |
| stefan dessloch | 2 | 91 | 11.69 |