This site uses an STK virtual tape system cluster, which accumulates many virtual tape volumes onto a disk array cache and then to multi-volume-cartridges loaded robotically. It also replicates those virtual tape volumes to a disaster recovery site where they are cached and a second robotic system loads them to a different library of multi-volume cartridges. I'm just becoming familiar with STK virtual tape and making some observations and have many questions.
First of all, real tape's update mode is primarily append. But using disk, once the virtual volume is written to a multi-volume cartridge, assuming it's not the last volume on the multi-volume cartridge, no physical straight-forward append seems possible. I don't think it chains virtual volume "extents" in different locations on the mulit-volume cartridge or across them to simulate a virtual append ??? I assume for a non-scratch ring-in virtual mount of a pre-existing volser it loads the original virtual volume data from the multi volume cartridge to its disk cache and then after the job finishes appending data it just marks the original data area free in its CDS directory and rewrites the original data with appended data in a new location on a multi-volume cartridge. So virtual tape probably handles applications which append data to old tapes rather poorly, fragmenting the data on the cartridges until background reclaim processing reorganizes them. Doesn't HSM L2 do a lot of append processing, adding to the end of tapes? This site doesn't currently replicate HSM L2 tape volumes nor sysout archive migrate or backup tape volumes to the disaster recovery site. I am considering changing that. Today I will quantify the increase in data traffic. But quantifying the increase in fragmentation and need for reclaim/reorg is more difficult. Should I even worry about it? With virtual tape meeting the same objective of using more of the entire physical tape, can HSM be run in a mode which does not append to its old tapes? That would use many more tpae volumes, but heck they're "virtual." How can I quantify the greatly increased number of volsers that would require in my CA-1 tape library? Is this even possible? ---------------------------------------------------------------------- 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
