Database Reference
In-Depth Information
•
Store Metrics
: This exposes the following metrics:
° Stores metrics: This provides the total number of storage files that
are spread across all regions managed by the region server
° File index size metric: This provides the sum of the block indexes
for all store files in MB
•
I/O metrics
: The region server provides the following metrics to track the
I/O performance in milliseconds:
° FS read latency: This provides the filesystem read performance,
that is, the time taken to load a block from HDFS
° FS write latency: This provides the filesystem write performance for
all writes, including the write-ahead log to the actual disk writing
° FS sync latency: This provides performance statics to sync the
write-ahead log records to the file system
• Other metrics: The region server also provides additional metrics for the total
number of read and write operations summed up for all available regions on
the region server. The framework also provides metrics for the actual request
rate per second and available regions within any region server.
JVM metrics
The primary aim of JVM metrics is to ine-tune the HBase performance. Every HBase
process collects and exposes suficient JVM-related details useful to ine tune the
server performance. The following are the different metrics that are exposed under
this category:
• Garbage collection: This provides the GC count and GC timing
• Memory: This provides the memory usage by Heap and JVM internal
• Thread: This provides the count of thread in different thread states
Info metrics
Apart from all the preceding metric categories, the HBase metric framework also
exposes common informational metrics that provide the ixed information about
the different HBase processes such as HBase build data, version, HDFS build date,
HDFS version, the user that built HBase or HDFS, and so on.
