nosql - Cassandra rack concept and database structure -


i newbie in cassandra , want know cassandra racks. , confuse cassandra structure. suppose have around 70 column family in cassandra , 2 aws2 instance.

1.how many data center used. 2.how many nodes each rack have ? 3.is possible divide column family in multiple key space.  

the intent of making cassandra aware of logical racks , data centers provide additional levels of fault tolerance. idea (as described in document, under "network topology strategy") application should still able function if 1 rack or data center goes dark. essentially, cassandra...

places replicas in same data center walking ring clockwise until reaching first node in rack. networktopologystrategy attempts place replicas on distinct racks because nodes in same rack (or similar physical grouping) fail @ same time due power, cooling, or network issues.

in way, can query data local_quorum, in quorum ((replication_factor / 2) + 1) computed nodes present in same data center coordinator node. reduces effects of inter-data center latency.

as questions:

  1. how many data centers used entirely you. if have 2 aws instances, putting them in different logical data centers possible, makes sense if planning use consistency level one. as-in, if 1 instance goes down, application needs worry finding 1 other replica. then, snitch can find data on 1 instance, or other.

  2. again, can define number of nodes wish have each rack. indicated #1, if have 2 instances, there isn't gained splitting them different data centers or racks.

  3. i not believe possible divide column family on multiple keyspaces. think know you're getting at. each keyspace created on each instance. have 2 instances, able specify replication factor of 1 or 2. if had 3 instances, set replication factor of 2, , if lost 1 instance still have access data. have 2 instances, need able handle 1 going dark, want make sure both instances have copy of every row (replication factor of 2).

really, logical datacenter/rack structure becomes more-useful number of nodes in cluster increases. two, there little gained splitting them additional logical barriers. more information, read through 2 docs linked above:

apache cassandra 2.0: data replication

apache cassandra 2.0: snitches


Comments

Popular posts from this blog

Android layout hidden on keyboard show -

google app engine - 403 Forbidden POST - Flask WTForms -

c - Why would PK11_GenerateRandom() return an error -8023? -