[OpenAFS-devel] Re: I/O parallelism: was: [AFS3-std] Re: [OpenAFS-devel] convergence of RxOSD, Extended Call Backs, Byte Range

[Matt W. Benjamin]

Oh, something I forgot for roman numeral I

I. 4.  Request-queuing/polling avoidance.

As this protocol will be serializing contending I/O operations originating at different clients, a (reliable) deferred operation mechanism should be defined to avoid client polling, and, where possible, ensure fairness.  I would propose that a mechanism similar to that proposed for deferred locks in recent byte-range locking drafts, with fall-back to polling, would resolve the issue.

Thanks!


Matt



----- "Matt W. Benjamin" <matt@linuxbox.com> wrote:

> 
> This note looks forward to the later mail in which the signatures for
> new begin/end IO routines are described e.g., RXAFS_StartAsyncFetch,
> as well as back to Tom's original list of issues.
> 


-- 

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. 734-761-4689
fax. 734-769-8938
cel. 734-216-5309