Moinak Ghosh wrote: > >>> The argument that disks are now so large that it's no longer >>> necessary to worry about space simply isn't true. Systems with >>> 18G drives (or similar) are still commonplace and entirely viable. >>> With the growth in virtualization, large drives will get chopped up >>> into much smaller chunks for allocation to installed systems. >>> And many Sun systems still get shipped with 73G drives. While >>> this may seem excessive now, it's not going to look generous >>> in 5 years time. Besides, Sun still sell reconditioned systems with >>> > > I think 5 years is an optimistic view. > 80GB Drives are not enough for me today! 120GB will not be enough in > 2 years time. > >>> 9G drives, and we mustn't completely ignore the hobbyist market >>> where less generous configuarrions are common. >>> >> >> >> If Sun wants to drop the SXCR program, then abandoning in-place >> upgrades would be one way to diminish the effectiveness to near >> zero. >> > > Disk space might be cheap, but enough software and data bloat is > present > and growing today to gobble up whatever space you can throw at it. IMHO > not supporting in-place upgrade will hit workstation users hard. > Most of the > disk space available will be partitioned off for various stuff and > there will > be multiboot configurations. It will be a pain to allocate space for > liveupgrade. > A short downtime is not a factor in this scenario. > Since live upgrade only deals with the BE the data bloat issue isn't an issue that is made worse by using live upgrade. The Solaris distribution certainly has grown in size but in comparison to the rate at which disks are growing it is a fraction of the total disk space available today.
Certainly, live upgrade requires additional planning when setting up a system. But, so does in place upgrade. You have to ensure your root slice is large enough to handle the upgrade, including future size changes in Solaris. If it isn't at any point in time you must resize this partition, beg and borrow from others to get the upgrade to succeed. The biggest roadblocks I see to not supporting in place upgrades for existing users are: 1. The current implementation of live ugprade is error prone and then handles the errors encountered badly 2. It is somewhat labor intensive to setup a live upgrade environment initially 3. Existing users will have to reconfigure their systems(with some pain) to get to the ability to do live upgrades. We can fix the first 2. We can help with 3 via some migration documentation, and providing tools that run on Solaris via opensolaris, like gparted, that help the users repartition their drives. We can't fix the need for additional disk space to host the ABE. But, I still argue that an additional 5GB for an ABE is small in comparison to the disk drives that are available today. > -> 2. There is no rollback mechanism, short of a full restore, if > something gets broken. > > ZFS root should help here. > > Community participation and participation of the individual > contributor in trying > out newer versions of Solaris on different hardware will diminish > greatly > without in-place upgrade. Why? If we can help them get to the live upgrade setup, and show them the benefits of using live upgrade? Keep in mind that the live upgrade we are talking about with Caiman will be much different(and improved) than the current live upgrade technology that we have today. thanks, sarah ****
