On Sunday 24 October 2010 00:08:11 Juiceman wrote:
> On Oct 23, 2010 1:01 PM, "Matthew Toseland"
> wrote:
> > We are still doing quite a lot of disk I/O, especially on writes. And a
> typical node does a lot of writes. The quadratic probing mechanism in the
> salted hash store means we have to do
On Sunday 24 October 2010 00:08:11 Juiceman wrote:
> On Oct 23, 2010 1:01 PM, "Matthew Toseland"
> wrote:
> > We are still doing quite a lot of disk I/O, especially on writes. And a
> typical node does a lot of writes. The quadratic probing mechanism in the
> salted hash store means we have to do
On Oct 23, 2010 1:01 PM, "Matthew Toseland"
wrote:
> We are still doing quite a lot of disk I/O, especially on writes. And a
typical node does a lot of writes. The quadratic probing mechanism in the
salted hash store means we have to do 5 seeks on a write - 10 on a write to
the cache, because we a
We are still doing quite a lot of disk I/O, especially on writes. And a typical
node does a lot of writes. The quadratic probing mechanism in the salted hash
store means we have to do 5 seeks on a write - 10 on a write to the cache,
because we also check the store. And counting bloom filters are
On Oct 23, 2010 1:01 PM, "Matthew Toseland"
wrote:
> We are still doing quite a lot of disk I/O, especially on writes. And a
typical node does a lot of writes. The quadratic probing mechanism in the
salted hash store means we have to do 5 seeks on a write - 10 on a write to
the cache, because we a
We are still doing quite a lot of disk I/O, especially on writes. And a typical
node does a lot of writes. The quadratic probing mechanism in the salted hash
store means we have to do 5 seeks on a write - 10 on a write to the cache,
because we also check the store. And counting bloom filters are
