[
https://issues.apache.org/jira/browse/HDFS-5442?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14121182#comment-14121182
]
Hari Sekhon commented on HDFS-5442:
-----------------------------------
Some thoughts I had on this:
1. Having multiple NameNodes at each DC is non-optional
2. WanDisco does multiple NameNodes at each DC with majority global quorum with
a weighting priority for the primary DC.
3. While zero replication is great, a single global namespace spanning multiple
datacenters would be awesome. One HDFS namespace globally with snapshots in
case anyone deletes anything, with data available for local read processing at
all sites.
4. Path-level replication control should be included to be able to exclude /tmp
and similar scratch directories to reduce Wan replication traffic
5. Async cross DC should be the default as otherwise performance will drop
through the floor.
6. Should have tunable consistency (think Cassandra), where you can choose to
have it confirm replication to 1 or more nodes at other DC before confirming
write.
7. Should have bandwidth control to not flood the wan link.
8. Must have a monitoring interface to determine the state of replication lag
of secondary DC.
9. Should support multiple datacenters using global quorum and replication of
blocks to all DCs.
10. Should replicate each unique block only once to a datanode at other DC,
which in turn replicates to other datanodes inside same DC, minimizing wan
traffic and relying on the in-built HDFS checksums.
10. Should support chain linking replication paths so 2nd DC replicates to 3rd
DC instead of primary DC having to send the same data twice through it's wan
link to both of them.
> Zero loss HDFS data replication for multiple datacenters
> --------------------------------------------------------
>
> Key: HDFS-5442
> URL: https://issues.apache.org/jira/browse/HDFS-5442
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Avik Dey
> Assignee: Dian Fu
> Attachments: Disaster Recovery Solution for Hadoop.pdf, Disaster
> Recovery Solution for Hadoop.pdf, Disaster Recovery Solution for Hadoop.pdf
>
>
> Hadoop is architected to operate efficiently at scale for normal hardware
> failures within a datacenter. Hadoop is not designed today to handle
> datacenter failures. Although HDFS is not designed for nor deployed in
> configurations spanning multiple datacenters, replicating data from one
> location to another is common practice for disaster recovery and global
> service availability. There are current solutions available for batch
> replication using data copy/export tools. However, while providing some
> backup capability for HDFS data, they do not provide the capability to
> recover all your HDFS data from a datacenter failure and be up and running
> again with a fully operational Hadoop cluster in another datacenter in a
> matter of minutes. For disaster recovery from a datacenter failure, we should
> provide a fully distributed, zero data loss, low latency, high throughput and
> secure HDFS data replication solution for multiple datacenter setup.
> Design and code for Phase-1 to follow soon.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
