On Fri, 27 Oct 2023 16:35:39 +0100, Jack Zukt <jzuk...@gmail.com> wrote:
>Is there anyone out there using cloud storage for backup or dataset >migration, using IBM Cloud Tape Connector for z/OS or Model9? I did a quick search and found an IBM presentation https://conferences.gse.org.uk/2019/agenda/presentation/1775 I took a quick look and verified that it requires using AMATERSE on the DFDSS file to retain the file attributes and blocking. While I don't have experience with this, there is a lot that we can infer and when problems may arise. Since Crayford worked for the original vendor, he should be able to tell us whether my assumptions about the internals are correct. ----- Using DFDSS volume backup and send it to the cloud ------ Run DFDSS backup -> AMATERSE pack -> copy to storage -> cloud put object (volume backup) -> delete work files and free storage. Considerations for backup: 1. Do you need encryption or compression for the file sent to the cloud object. 2. Backup to tape drives / VTS can have much larger block sizes than disk which means restore / recall may require these devices. 3. Identify resources required for each backup running in parallel. If you use disk, then potentially each backup could require 2 disks. Similar considerations when using tape or VTS. Backup time will increase which may require scheduling changes. 4. Ensure a unique cloud object names for each version of a volume is retained (e.g. GDG). 5. Most likely the cloud object must be sent from storage and be large enough to contain the entire backup. -------- Using HSM to recall / restore file or volume ---------- Recall / restore request -> cloud get object (volume backup) to storage -> move storage to disk -> AMATERSE unpack -> restore file / volume and delete work files. 1. Considerations for recall / restore a file or a volume are the same as backup except the steps are reversed. 2. Recall / restore is unplanned. Look at your recall history to see if you can handle the worst case. Are you affected by recalls after vacations? Are you affected by yearly processing? Are there other factors that could increase recall / restore volume? 3. Disaster recovery must be completely reworked for using a public cloud like Amazon. Using real tape meant you aren't affected by other departments also recovering from the disaster. Cloud recovery means that you could be competing for a critical resource when others inside and outside of your company. Imagine an earthquake in Los Angeles where you potentially use the same recovery services. 4. Do you only restore the OS and require various departments to HRESTORE their files? While running concurrently, will this over burden your disk, tape and storage usage during an extreme and unusual situation when people are under extreme pressure? Will these restores be inefficient because the same backup volume is downloaded multiple times. Using an IBM TS7700 VTS would certainly be far less painful and more transparent than this product. Any company can make this work but it will require commitment to review, revise and change their processes. This is not a simple change of medium. Ignoring the implications and impacts will cause major grief when you can least tolerate it. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN