Hello, fio folks. I'm working on simulating workloads for SAS and am having a bit of trouble getting the read and write mix to model how SAS works. In my example, SAS will read in two data sets from separate OS files, then write a merged data set out to a new OS file. The issue is that SAS will read each data set one line at a time, then write the appropriate lines out to the new file before reading in the next lines from the source data sets. I've tried a few different fio job configs and can't seem to get it to work how I expect. This seems to make sense: [global] disable_clat=1 disable_slat=1 clat_percentiles=0 direct=1 directory=/sasdata1 ## Simulates interleaving of two data sets ## [source_data_sets] rw=read filesize=100m nrfiles=2 [destination_data_set] rw=write size=200m But the actual behavior is that the second job doing the writing seems to block the first job from reading and the reads only really proceed after the write job is finished. I've also tried a config with a single job using mixed sequential read/write, and while that may end up being good enough, it seems like it doesn't necessarily match what I'm trying to simulate. Basically I want the first job to do 2 reads, then the second job to do a write and continue in that roundrobin kind of way which is how SAS would process the files. I've read through the HOWTO a few times and think I'm probably just not understanding what a couple options here or there could do for me. Thanks for your help! Regards, Spencer -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
