Databases Reference
In-Depth Information
FIGURE A.1
The NetApp AutoSupport team deployed a NetApp Open Solution for Hadoop, which includes Hadoop
clusters on NetApp E2600 storage systems and a NetApp FAS2040 system running Data ONTAP 8.
point of failure common in traditional Hadoop-clustered deployments and instead offers full redun-
dancy and automated path failover, along with online administration (
Figure A.1
).
This highly efficient CDH processing will enable our AutoSupport team to quickly mine hundreds of
terabytes of data, for real-time analysis of customer system event and performance data and rapid
resolution of any issues.
—Marty Mayer, director of NatApp AutoSupport
Impact: high performance for big bandwidth applications
AutoSupport predicts significant performance gains running on the NetApp Open Solution for
Hadoop. By supporting even the most bandwidth-intensive applications, the solution will enable
the NetApp AutoSupport team to meet stringent SLAs for parsing and loading data. In one case,
AutoSupport wanted to correlate disk latency when a disk was hot with the type of manufacturer
disk to identify whether there was a relationship between the two. The report requires a query of
24 billion records, which took four weeks to run on the incumbent environment. On the massively
parallel NetApp Open Solution for Hadoop, that query returns in 10.5 hours. That's a 64 times query
performance improvement. Mayer noted,
The productivity and customer service benefits enabled by the NetApp solution are significant.
Running the NetApp Open Solution for Hadoop gives us the ability to turn an unwieldy data explosion
into a highly manageable environment. It also will allow us to perform deeper analytics than before,
which will provide better monitoring and troubleshooting of NetApp customer storage systems.
