> Another observation: when you have to recycle a vnode, you get one that's
> basically chosen at random with regard to the current head position.  This
> is REALLY BAD.  To fix it, we'd need to index the vnode cache some other
> way than as a hash table, but the fix might give a really large performance
> increase in some cases.

I'm pretty sure I suggested this... what, 6 years ago?

One thing to remember is that the more ways you index things, the
harder it is to parallelize the file system code -- and those 128-way
NUMA machines are pretty sexy.