> Frank's result based on testing of fsal vfs can't be generalized to > configurations which actually have latency, so, no?
This really needs to be repeated with other FSALs, and probably also it would be best for the client to at least be a separate VM if not a separate physical machine. Of course that might increase the network latency between client and server, but then if that dominates, caching or not might make no difference. To Swen's question, this was not on a directory being thrashed. That should be dumping cache, so then the question would be what is the impact of loading up the cache vs. using uncached readdir. Likely uncached would have the best performance on a heavily thrashing directory with 200k files if the client just steadfastly marches through the READDIR since we wouldn't stop to read more entries from filesystem than an individual client request consumes (ok, right now, we read one more entry than we consume, I hope we can fix that, or at least reduce how often we do that). Frank --- This email has been checked for viruses by Avast antivirus software. https://www.avast.com/antivirus ------------------------------------------------------------------------------ Developer Access Program for Intel Xeon Phi Processors Access to Intel Xeon Phi processor-based developer platforms. With one year of Intel Parallel Studio XE. Training and support from Colfax. Order your platform today. http://sdm.link/xeonphi _______________________________________________ Nfs-ganesha-devel mailing list Nfs-ganesha-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs-ganesha-devel
