On Wed, 8 Jul 2009 19:39:03 -0500, Dan <[email protected]> wrote:
>Sorry to take this thread on a side road so early but ... >Wayne, this may benefit you as you may decide to rethink using VTS. > >I've see quite a few sites mention that they use VTS (or other virtuals) to >manage their ML2 volumes. Is this a realistic thing to do? > >To perform a single dataset recall you must first stage the entire ML2 >volume to cache. This may cause other files to be bumped from cache as the >HSM migration volumes are usually FULL volumes. > >Only after the volume is successfully staged may HSM begin the recall. At a >minimum, this will take twice the I/O traffic, plus it may impact what is >currently resident in the VTS cache. > >If you have a virtual library you more than likely have an automated >library. Why not simply use the automated library for ML2? > >Dan Another detail ... FSM for HSM ( fast subsequent migration) is not working for multivolume.( I don't think it has changed) . So if you recall a multi volume tape dataset and migrate it again , the migration will have to happen physically.( instead of just setting the control bits) Bruno Sugliani zxnetconsult(at)free(dot)fr http://zxnetconsult.free.fr ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
