hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Doug Cutting (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-692) Rack-aware Replica Placement
Date Fri, 10 Nov 2006 19:10:38 GMT
    [ http://issues.apache.org/jira/browse/HADOOP-692?page=comments#action_12448822 ] 
Doug Cutting commented on HADOOP-692:

We need an extensible API for describing network topology & distance.  Here's stab at

public Interface NetworkLocatable {
  /** Return the names of each network level of this node: host, rack, datacenter, etc. */
  String[] getNetworkLocation();

It is tempting to use something more abstract, e.g. a distance method, but that would not
provide a concise indication of the topology, and would require linear scans, or storage of
a full distance matrix, when, e.g., placing blocks.

As a first implementation, I like the idea of trying to distribute blocks to increasing network
levels: first to the same host, then to the same rack, then to the same datacenter, etc. until
the replication level is exhausted.  If the replication level is higher than the number of
levels, then a uniform distribution at the highest level should be sought.

Down the road we may wish to make this strategy extensible, or to have a more elaborate distance
calculation, but we can first probably get a lot of milage from this simple model.

> Rack-aware Replica Placement
> ----------------------------
>                 Key: HADOOP-692
>                 URL: http://issues.apache.org/jira/browse/HADOOP-692
>             Project: Hadoop
>          Issue Type: Improvement
>          Components: dfs
>    Affects Versions: 0.8.0
>            Reporter: Hairong Kuang
>         Assigned To: Hairong Kuang
>             Fix For: 0.9.0
> This issue assumes that HDFS runs on a cluster of computers that spread across many racks.
Communication between two nodes on different racks needs to go through switches. Bandwidth
in/out of a rack may be less than the total bandwidth of machines in the rack. The purpose
of rack-aware replica placement is to improve data reliability, availability, and network
bandwidth utilization. The basic idea is that each data node determines to which rack it belongs
at the startup time and notifies the name node of the rack id upon registration. The name
node maintains a rackid-to-datanode map and tries to place replicas across racks.

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message