Hello. Please pardon this post from someone new to the list if this is irrelevant, but I've recently been involved in setting up iscsi on a performance testing cluster, and I have observed that a single tgtd process does seem to be a bottleneck in a high-throughput (10GbE) environment, and that performance improves when I use multiple tgtd processes. If you want the details about this configuration, please let me know. Regards -- Steve On 2010-04-05 22:48, FUJITA Tomonori wrote: > On Tue, 6 Apr 2010 15:35:25 +1000 > ronnie sahlberg<ronniesahlberg at gmail.com> wrote: > > >> I dont think that is viable for Chandra. >> >> TGTD is pretty limited performance wise with its central eventloop. On >> a many-core machine with a lot of memory >> the central event loop quickly becomes a bottleneck. >> > Can you have the performance analysis with tools such as perf? > > The most of network processing happens in kernel space. The I/O > processing too. They are not related with the number of user > processes. > > Yeah, tgtd would be a bottleneck but there are lots of other possible > bottlenecks. > > > >> For these usecases you really want to run multiple instances of tgtd. >> In particular if you have a high-end storage subsystem attached >> that can deliver high enough throughput. >> >> Perhaps have one instance of TGTD for each 10GbE adapter. Perhaps one >> instance of TGTD for every 2 cores. >> > As I said yesterday, I think that we need to think about using > multiple threads for targets (or initiators) if necessary. But we need > the evidence first. > -- > To unsubscribe from this list: send the line "unsubscribe stgt" in > the body of a message tomajordomo at vger.kernel.org > More majordomo info athttp://vger.kernel.org/majordomo-info.html > -- To unsubscribe from this list: send the line "unsubscribe stgt" in the body of a message to majordomo at vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html |