<div dir="ltr">Hi everyone,
<div><br></div><div>I'm not sure, whether the lustre or the MPI forum is the right place for my question.</div><div><br></div><div>The question is about the ROMIO optimization on Lustre,</div><div>In one SC'08 paper,</div>
<div><a href="https://mail.ttu.edu/owa/redir.aspx?C=yUmbVUH4hUWLFEWFA2GcoiKOEhnhitAIatZfGT92-aN2MTXitjDjPgfE9EfJkJF9q3XAaOQ_iME.&URL=http%3a%2f%2fusers.eecs.northwestern.edu%2f%7ewkliao%2fPAPERS%2ffd_sc08_revised.pdf" target="_blank" style="font-family:monospace;font-size:16px">http://users.eecs.northwestern.edu/~wkliao/PAPERS/fd_sc08_revised.pdf</a><br>
</div><div>, it's said that the way ROMIO assigns the file domains to I/O aggregators will not make two aggregators access the same OST<font color="#000000" face="monospace" size="3">.</font></div><div><font color="#000000" face="monospace" size="3"><br>
</font></div><div><font color="#000000" face="monospace" size="3">In my understanding, this means, the data locality on Lustre layer has been taken care of in the ROMIO, such that the aggregators will not have competition on the same OST.</font></div>
<div><font color="#000000" face="monospace" size="3"><br></font></div><div><font color="#000000" face="monospace" size="3">My question is "is this optimization used in all current lustre system, e.g., Hopper at NERSC?"</font></div>
<div><font color="#000000" face="monospace" size="3"><br></font></div><div>Thanks,</div><div><br></div><div>Jailin Liu</div><div>Ph. D student </div><div>Texas Tech University</div></div>