Database Reference
In-Depth Information
Memttablle
An in-memory representation of data that has been recently written. Once the memtable is
full, it is flushed to disk as an
SSTable
.
Merklle T
e Tree
Perhaps better known as a “hash tree,” a Merkle tree is a binary tree data structure that sum-
marizes in short form the data in a larger dataset. In a hash tree, the leaves are the data blocks
(typically files on a filesystem) to be summarized. Every parent node in the tree is a hash of
its direct child node, which tightly compacts the summary.
In
Cassandra,
the
Merkle
tree
is
implemented
in
the
org.apache.cassandra.utils.MerkleTree
class.
Merkle trees are used in Cassandra to ensure that the peer-to-peer network of nodes receives
data blocks unaltered and unharmed. They are used in cryptography as well to verify the con-
tents of files and transmissions, and are used in the Google Wave product. They are named
for their inventor, Ralph Merkle.
Mullttiigett
Query by column name for a set of keys.
Mullttiiget S
t Slliice
Query to get a subset of columns for a set of keys.
Node
An instance of Cassandra. Typically a Cassandra cluster will have many nodes, sometimes
collectively called the node ring, or just “the ring.” A node refers to any Cassandra server in
a cluster, whereas “replica” refers to a node that specifically has a copy of some data from
another node.
Node T
e Tooll
This is an executable file with the path bin/nodetoolthat inspects a cluster to determine
whether it is properly configured and performs a variety of maintenance operations. The
commands available on
nodetool
are
cleanup
,
clearsnapshot
,
compact
,
cfstats
,
de-
commission
,
drain
,
flush
,
info
,
loadbalance
,
move
,
repair
,
ring
,
snapshot [snap-
shotname]
,
removetoken
, and
tpstats
.
For example, you can use
nodetool drain
to prevent the commit log from accepting any
new writes.
