Re: The 2GB chunk size limit and Sparc storage arrays

In article <4dln51$gof@fiesta.srl.ford.com> Mike Reed - Radiix <mreed> writes: > >I (think) I've come across an irritation while using the >Sparc storage array with Online Disk Suite 4.0. Our Sun SE >told me (and showed me) that the maximum number of logical >names that can be used on a system is 127 (the /dev/md/rdsk >names). If this is true, ... This is not really true. Out of the box, SDS4.0 has a default of 127 possible metadevices (d0 - d126), but this can be changed by adding an entry to /etc/system which will allow you to create as many metadevices as you want; the only thing to note is that if you change the parameter in /etc/system then you have to scale the size of your MetaDB replicas accordingly. The default size for the replicas is 517k for each replica, and you need at least 3 (but preferrably not all on the same controller). If you increase the number of possible metadevices, you must allow for much larger MetaDB replicas. If you check the SDS4.0 AnswerBook, you can read about it in there. We were not using Solstice DiskSuite with our SPARCStorage Arrays originally, but now we are. The mirroring is a great feature, but aside from that, with the fast_write cache enabled and striping across multiple drives, the perfor- mance is outrageous! I was happy with the performance of just the individual raw drives, but after reconfiguration with SDS4.0 I couldn't believe the difference; and that GUI tool is the greatest, no more editing the md.tab file! I would highly recommend it, especially if you want to mirror your disk or use a higher level RAID configuration. Keith.