HBase doesn't know all data are in the block cache. it had to look at
HTable firstly to get "block_id"(tablename + offset), then find it in the
so if all data in the block cache, you just avoid to read data from hfile
directly, save some I/O time. but it depends on your data size.
if you always return a large data size, the read performance cannot be
On Mon, Jun 3, 2013 at 5:13 PM, ramkrishna vasudevan <
[EMAIL PROTECTED]> wrote:
> What is that you are observing now?
> On Mon, Jun 3, 2013 at 2:00 PM, Liu, Raymond <[EMAIL PROTECTED]>
> > Hi
> > If all the data is already in RS blockcache.
> > Then what's the typical scan latency for scan a few rows from a
> > say several GB table ( with dozens of regions ) on a small cluster with
> > 4 RS ?
> > A few ms? Tens of ms? Or more?
> > Best Regards,
> > Raymond Liu