I think this is expected. It's the first time I tested this in 0.96 or later.
Still +1 here.
I'll do the same test in 0.96 next week. I'd expect it'll be similar to 0.98.
With "pure" scan performance I mostly mean the internal "friction". I.e. everything in the blockcache and all KVs filtered at the server with a filter.
But as you said, this is not a discussion thread.
We can tackle performance improvements later in the cycle.
From: Andrew Purtell <[EMAIL PROTECTED]>
To: "[EMAIL PROTECTED]" <[EMAIL PROTECTED]>; lars hofhansl <[EMAIL PROTECTED]>
Cc: "[EMAIL PROTECTED]" <[EMAIL PROTECTED]>
Sent: Saturday, January 25, 2014 3:13 PM
Subject: Re: [VOTE] The 1st HBase 0.98.0 release candidate is available for download
On Sat, Jan 25, 2014 at 2:35 PM, lars hofhansl <[EMAIL PROTECTED]> wrote:
Raw scanning is about 30% slower than 0.94 in pure scan performance.
My initial reaction to that is -0, but I don't think it holds up .0, just charts out work needed for .1. When developing on trunk we looked at the deltas between trunk and trunk+changes, not between releases. How different is this from 0.96 I wonder.
Problems worthy of attack prove their worth by hitting back. - Piet Hein (via Tom White)