If each region is 4G, and there are 200 regions in one regionserver, a regionserver will support 800G hbase data. Is it too small?
2011/11/10 Doug Meil <[email protected]> > > http://hbase.apache.org/book.html#perf.number.of.regions > > > > > > > On 11/9/11 9:25 PM, "吕鹏" <[email protected]> wrote: > > >Thanks a lot for you help. > > > >But for question 2 and 3, the Apache book does not get give a direct > >answer, specially question 2. > > > > > >2011/11/10 Doug Meil <[email protected]> > > > >> > >> Hi there- > >> > >> re: #1 > >> > >> http://hbase.apache.org/book.html#perf.configurations > >> > >> Regarding this, for the 0.90.x codebase the largest recommended region > >> size is 4Gb. The "20Gb" number in the book now was for a cluster > >>running > >> which was running Hfile v2 format, which is different than what is in > >> 0.90.x. I will update the book this weekend. > >> > >> re: #2 and #3 > >> > >> See http://hbase.apache.org/book.html#rowkey.design > >> > >> > >> > >> On 11/9/11 1:32 AM, "吕鹏" <[email protected]> wrote: > >> > >> >I have three question: > >> >1 How large a region will be, if i want to use map-reduce to analyze > >>the > >> >data in hbase table. > >> >2 How many region in a region server is suitable? > >> >3 If the rowkey in my application is not dispersive such as md5 or > >>hash, > >> >how can i control the hotspot of region split? > >> > > >> >thx > >> > >> > >
