CA Disk determines the amount of space to allocate (the amount of space it needs on a diskpool volume) in the following manner. It assumes that quite frequently the primary copy is on disk and the duplicate is on tape or a 3480 cartridge. As stated earlier, the target volume leads CA Disk to the device type, which in turn dictates the blocksize to be used. This, coupled with the density and length of tape being used, allows an easy and very accurate calculation of the tape capacity. (Because the ends of 3480 tapes cannot be clipped off, the most common variable in the calculation has been removed.)
The optimal disk allocation is the exact same capacity as its backup medium, such that when one is filled and both must be closed, nothing is wasted on the other medium. To do this on an exact equivalence, however, would require very large amounts of free space to be available. Therefore, CA Disk selects 1/16th of that value as default primary and secondary space allocations. In many cases, an archive run does not need a full tape, and one or two extents are more than enough space. However, when a large archive run is made and the diskpool volumes contain plenty of free space, CA Disk obtains as many as possible of the 16 extents before closing the archive data set. This maintains high tape usage as well. Any excess (unused) disk space is, of course, released immediately when the data set is closed.
CA Disk also considers the size of the data sets being archived when determining how much space to allocate. If the first data set to be archived is larger than the default primary space, the primary space is reset to the size of the data set. If free space equal to this new primary value is not available but the default value is, CA Disk attempts to archive the input data set to a multivolume output data set. This is done only as a last resort and you should not use a Storage Class that is defined with guaranteed space because unused space is only released in the active volume at CLOSE time (there is a possibility that all preallocated volumes would not be used because the data is written in compressed format).
Once the target archive data set is allocated and the first data set to be archived is copied to it, CA Disk continues to copy additional data sets (the second through nth) as long as there is sufficient space to hold them. To determine whether sufficient space is available, CA Disk examines the unused space in the current extent, and checks to see if additional extents can be obtained. If there is not sufficient space, the current disk archive data set is closed and a new one is allocated (like swapping to a new tape).
If your analysis or practical experience indicates that the default method of calculating space allocations should be changed, specify sysparms SPACEPRI and SPACESEC with a three-digit number representing the number of megabytes of disk space to allocate for each.
When archiving to disk, you normally want the archive data sets to be large, as restoring from a large data set is as fast, or faster, than restoring from several small data sets. CA Disk provides options for balancing the need to conserve storage space with the need to keep overhead to a minimum. If you set the primary space value (sysparm SPACEPRI) to the size you want to use for staging, then set the secondary space value (sysparm SPACESEC) to 000, CA Disk creates smaller archive data sets, but more of them.
|
Copyright © 2015 CA Technologies.
All rights reserved.
|
|