Database Reference
In-Depth Information
Figure 8-8
.
An image in JConsole captured during the run of read mode
For more details on how to configure and connect JConsole with Cassandra you
may refer to
Chapter 1
.
For third-party tools, DataStax's OpsCenter provides detailed information about
running the Cassandra cluster. For more details on usage of the DataStax OpsCenter
please refer to
www.datastax.com/what-we-offer/products-ser-
Compaction Strategy
During Cassandra stress testing, we have seen that writes in Cassandra are blazingly
fast. The reasons behind this are that there are no updates and all columns are directly
flushed into sstables. This means that at some point there would be many versions of
column values which may be present in form of multiple sstables. Similarly physical
deletes would not happen in real time but those are logically deleted instantly. Such
rows are called
tombstone
or
obsolete rows
. The process to remove such columns and
free some space at the server side is known as
compaction strategy
.
Compaction always runs in the background with async mode. Cassandra offers two
types of compaction strategies:
•
Size-tiered compaction strategy
