Database Reference
In-Depth Information
Summary
We know how to move data around and manage Cassandra instances to handle the produc-
tion situations. Cassandra provides simple one-liner commands to perform various com-
plicated maintenance tasks to make life easy. Scaling out, scaling in, removing live or dead
nodes, and load balancing are pleasantly simple and can be automated based on your con-
figuration using scripts. Backups fall in the same category, but restoration can be a bit
tricky. With the Cassandra data model, built-in replication, and support for multiple data
center setups, one can configure Cassandra such that it may never need to be backed up.
Also, for really large databases, it may be impractical to siphon out data instead of using
replication. It may make sense to back up in a case where the database is not very large and
one uses a replication factor of 1. In such a case, even a single malfunctioning server can
cause loss of data. However, RF=1 is a bad idea in production setup and backup can just re-
store the data until the latest backup is made.
OpsCenter gives a decent option to move on from the nodetool-based mechanism to more
sophisticated tooling. If you are just starting out or have a small cluster, it may not be worth
the effort it takes to configure OpsCenter. It is a good suggestion for large cluster owners.
Also, if you are just starting to look into Cassandra, it is more fun to play with CLI tools
such as
nodetools
and to check
netstat
,
top
, memory usage, and so on to see what
is actually happening underneath.
There is another Cassandra project for DevOps from NetFlix called Priam (
ht-
tps://github.com/Netflix/Priam/wiki
). The reason why Priam wasn't discussed in the
chapter is that it seems a bit outdated and confusing. Moreover, Priam is built to support
the AWS infrastructure and tied to the AWS configuration. So, it might not be useful to
everyone. If you decide to explore Priam, make sure you read this wiki entry first:
ht-
This chapter gives you enough knowledge to tackle an infrastructure issue. Now, as an op-
erations person, you need to keep a watch on how things are going without losing your
sleep and waking up from nightmares. The next chapter will walk you through the various
ways to monitor and recognize problems and troubleshoot the Cassandra infrastructure in
detail.
