In 0.20 its a global setting right now. We hope to support in memory column
families soon.
What do you think the appropriate setting should be? My thought is to avoid
block cache on very large blocks and otherwise let the lru do its job.
On Jun 4, 2009 11:25 PM, "Billy Pearson" <sales@pearsonwholesale.com> wrote:
I created a table with one column in my cluster and started putting data in
to it I noticed that even with blockcache = false it still uses the block
cache
Is this a know problem or is there plans to remove the option from the table
create alter commands?
Billy
|