Cloud with IBM i (AS/400): Can I save my data THAT way?

Your data is one of your most valuable assets, and ensuring its safety should be a top priority.

Whether you're already using cloud solutions or just starting to explore them, this article will provide you with actionable insights to enhance your backup practices.

Yes, I have heard (I am sure you have heard them, too) really interesting ideas about saving IBM i backups in different ways, when running on-premise and now the cloud-boosted ideas:

1. I want to save my data to a Pen-Drive/USB 3.1 ultra-fast external disk so you can download in the cloud
2. I want to give you our tape cartridges so you can restore them into our cloud instances
3. I have a NAS attached to the same network, so I can backup my data as I do with my x86 servers.
4. I want to create SAVE FILES and FTP to our VM running in XYZ Cloud/an Object Storage
5. I can take my backups using SAVE FILES/Image Catalog media files (Tape or Optical) and upload those files to my remote VM, so I can save those files using my Windows backup software in a Backup Server/Physical Tape/Object Storage
6. Why should I save my data? This system is so reliable/indestructible "the roof could fall on it, and nothing would happen."
7. We can save to a Virtual Tape/Optical Image Catalog using a NFS mounted Linux filesystem, so our backups will avoid using our expensive IBM Power disks, and then upload to XYZ Cloud.
8. I can take multiple snapshots/exports/templates when working on the cloud

IS THIS REALISTIC?

Short answer? Most of them are ideas that work, but need some polish in a real world scenario.

Let see...

1. PenDrives and USB Drives: You can attach PenDrives, not USB Drives. A Pendrive is considered an "optical" media device, and writing data to this media is slow.
2. When working in a cloud environment you lose physical access to your systems, but you can use a service provider to convert that physical media into virtual media. ie: We provide that kind of solutions to our customers...
3. You need specific protocols and releases to access NAS devices, but is possible. We can do that. Just remember: Cheap NAS devices are slow, and CPU, memory, disk technology (HDD or SSD), disk bays quantity and port link speed matters.
4. You can backup your data to SAVE FILES or IMAGE CATALOG files, and then upload to a file server. If you plan to use a second backup layer to move the media to a second repository (Object Storage, FTP Server, Backup Server), try to avoid that. Restoring a file could take forever.
5. Similar to point #4.
6. I think we don't need to discuss this point. If you think like this... think again and go back to your college books, or evoque the day you lost all your PC data because of an error or a virus. THERE IS NO AUTOMATIC BACKUP FEATURE AS ON VPS OR WEB SERVERS WHEN WORKING WITH IBM i, AIX, OR ANY OTHER ENTERPRISE WORKLOAD.
7. This sounds great, specially for small amounts of data to save, but you need a FAST LOCAL area network. Forget about using a remote server. And media should be rotated on every backup, something needs to run on the remote server to exchange media. The good news: Now you can compress the data before saving to this kind of media. Please, read this IBM page : "Remote Virtual Image Catalog"
8. You can use SNAPSHOTs before a change management event so you can rollback to a previous state. EXPORTs or TEMPLATEs are great mechanisms to save the entire LPAR, some kind of Cloud Option 21, but some limits apply, and you won't be able to restore just one file.

CONSIDERATION AND CONCLUSIONS

You need to think on restore procedure, RTO and RPO when backing up a system.

There is no magic solution, and sometimes you need to rethink all your networking before moving to the cloud.

VTL: You can use Falconstor VTL to emulate a Tape Library and backup your data with your well known tools, but remember, you won't be able to extract physically those cartridges, unless you sync to an on-premise facility.

Automation is the key

BRMS is a great tool. I really don't like it, but works. You can save to ICC (I don't feel confortable with ICC, sometimes just fail) or to a VTL. The great part with BRMS is the cleaning part. You can delete old media and avoid paying for old backups stored daily with no control.

If you like to script, you can do that. You an create a nice BASH script or CL to save data on your disk, compress and then upload to a Cloud Object Storage. But think carefully on how you will clean old media files and how long your restores will take.

A SAVE FILE per library sounds nice at first sight, but think of those huge libraries with more than 500GB of data. You need to download that monster file before restore. An Image Catalog allows to save data in fixed chunks.

You also need to consider the extra disk space for those backups.

Finally: External bandwidth can cost you (a lot of) money. Think carefully the right path for your backups. Backups can move a huge amount of gigabytes, and you could be paying the GBs you transfer. Look for flat rate file transfer, when possible. Same applies to replication solutions.

My advise: Think carefully and test before putting your new ideas in production. Paying an expert consultant could save you (a lot of) money.

Hope these can help you take the right decision for your IBM i backups when working in a cloud instance.

I suggest to read the "IBM Power Systems Virtual Server Guide for IBM i" and the "IBM i Backups with IBM Power Virtual Server".

If you plan to use an iSCSI VTL: "IBM i Removable Media: Support for iSCSI VTL".

As always, you can leave a comment or send me a message.

Good luck!!!

Diego E. KESSELMAN