I don't like it either, but I couldn't find another way to solve the
problem. I guess the question we need to ask is how many transactions will
actually lock something. Certainly no read transactions. Can you suggest
a good multi-thread concurrent server benchmark?
On Mon, Mar 11, 2013 at 4:08 PM, Nathan Williams <email address hidden>wrote:
I don't like it either, but I couldn't find another way to solve the
problem. I guess the question we need to ask is how many transactions will
actually lock something. Certainly no read transactions. Can you suggest
a good multi-thread concurrent server benchmark?
On Mon, Mar 11, 2013 at 4:08 PM, Nathan Williams <email address hidden>wrote:
> Making transactions that perform writes slower by doing the cache update /code.launchpad .net/~pbeaman/ akiban- persistit/ fix-1126868- lock-table- pruning/ +merge/ 152063
> synchronously seems like a bad conscious choice to make. Bucket locking,
> which recomputing the cache does, is exclusive right now. I have a feeling
> this will really hurt concurrency.
>
> Do we have any perf tests for this or am I being too pessimistic?
> --
>
> https:/
> You are the owner of
> lp:~pbeaman/akiban-persistit/fix-1126868-lock-table-pruning.
>