Just to add on to Ted's comment, ROOT has been dropped. META locations are now directly stored in ZK. See this ticket<https://issues.apache.org/jira/browse/HBASE-3171>for more.
Warm Regards, Tariq cloudfront.blogspot.com On Wed, Feb 26, 2014 at 12:54 AM, Ted Yu <yuzhih...@gmail.com> wrote: > You may have read http://hbase.apache.org/book.html#perf.reading > > For BlockCache, see this blog: > http://www.n10k.com/blog/blockcache-101/ > > Cheers > > > On Tue, Feb 25, 2014 at 11:03 AM, Upendra Yadav <upendra1...@gmail.com > >wrote: > > > I have go through no. of documents/slides about HBase read/get > operation... > > But I have lot of broken links and question to clear view about how hbase > > read operation happen... > > > > Like.... > > Client Communicate to zookeeper and get region server where -ROOT- table > > region reside...(And this is in memory) > > > > Then getting region server location for .META table region (And this is > in > > memory) > > > > Then location of region server where region of user table reside by > > checking start and end rowkey with given GET operation rowkey... > > > > then will search 1st in memstore.... if found then return data > > else > > BlockCache.... if not then BloomFilter if negative then return not > > found.... else > > > > read from hfile....(how i don't know.. please help me) > > > > Bloomfilter and BlockCache update and return data... > > > > Is this correct... Please suggest me some good document... > > > > Still i have doubt ... how BlockCache LRU works... > > >