hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Charles Mason" <charlie....@gmail.com>
Subject Re: [LIKELY JUNK]Back Up Strategies
Date Mon, 22 Sep 2008 19:35:48 GMT
On Mon, Sep 22, 2008 at 8:20 PM, Ding, Hui <hui.ding@sap.com> wrote:
> This should be something the operators of your data store worriy about.
> E.g., say hdfs uses three replicas, one should be on a local rack, the
> other on a different rack (to protect against power outage)
> And a third on a remote data center...
> If you have only a small cluster, then maybe use ups to guard against
> power outage and watch out for storms?
> After all, what are the chances that a meteorite hit your data center?
Well knowing my luck :)

There are some times when moving data from one cluster to another is
important. Moving data from a development cluster to a production one
is another useful feature.

I know some clusters are so vast its not so practical or not so
important depending on what the data represents.

Charlie M

> -----Original Message-----
> From: Charles Mason [mailto:charlie.mas@gmail.com]
> Sent: Monday, September 22, 2008 12:13 PM
> To: hbase-user@hadoop.apache.org
> Subject: [LIKELY JUNK]Back Up Strategies
> Hi All,
> I was wondering what the options there are for backup and dumping an
> HBase database. I appreciate that having it run on top of a HDFS
> cluster can protect against individual node failure. However that
> still doesn't protect against the massive but thankfully rare
> disasters which take out whole server racks, fire, floods, etc...
> As far as I can tell there are two options:
> 1, Scan each table and dump the entire row to some external location,
> like MySQL Dump does for MySQL. Then to recover simply put the new
> data back. I am sure the performance of this is going to be fairly
> bad.
> 2, Image the data stored on the HDFS cluster. Aren't there some big
> issues with it not grabbing a consistent image as some updates won't
> be flushed? Is there any way to force that, or to make it be
> consistent some way, perhaps via snapshoting?
> Have I missed anything? Anyone got any suggestions?
> Charlie M

View raw message