Virtualization Technology News and Information
Article
RSS
Faster than VTL: New Performance Horizons for Mainframe Backup and Archive Performance

By Gil Peleg, Founder & CEO of Model9

Mainframe backup and archive processes have a well established cadence that uses either physical tape libraries or virtual tape libraries (VTLs). However, in a modern data environment, VTLs are no longer the fastest option available. Today, new, cloud-based mainframe data management solutions are upending convention and redefining what is possible.

Newly Emerging Best Practices

Like any other reliable and familiar system, organizations have been reluctant to make changes in this routine, often because alternatives have seemed too complex. It's been the old mantra: "If it isn't broken, don't fix it." 

Now, though, the availability of simpler and more affordable technologies is leading some mainframe organizations on a new path to simpler, faster backup and archiving that not only reduces their compute overhead, it also allows them to better leverage data to achieve business wins with measurable bottom-line benefits. With outstanding metrics and substantial cost advantages, this approach clearly represents an emerging best practice for mainframe organizations.

This breakthrough results in increased ingestion throughput, and it is made possible by leveraging zIIP engines instead of mainframe MIPS and employing multi-agent load balancing to share the workload across the entire mainframe environment. It also involves both writing multiple data sets and breaking larger data sets into easier to handle "chunks," implementing parallelism through the use of multiple TCP/IP streams and compressing data with the help of zIIP or zEDC capabilities when available. 

The net result is to effectively increase available bandwidth and throughput. A perfect example of this is two separate banks, one US-based and the other in EMEA, who each saw huge backup and archive performance process gains when they used the technology.

The US bank was operating Oracle StorageTek Virtual Storage Manager (VSM) System to write to on-prem tape storage. Their goal was to cut back the large time windows required to complete backups. The new approach yielded a 20x improvement in full volume dump backup (reducing required time from 24 hours to just over one hour). Similarly, ingestion throughput went from 27MB/s to 547.96 MB/s.  Data set backups saw equally striking improvements, shrinking elapsed time from 5 min 34 seconds to just 36 seconds. Finally, DB2 recovery times were also sharply accelerated.

The EMEA bank also saw stunning improvements. They sought to replace IBM virtual tape libraries and associated software, increase throughput, and write directly to IBM Cloud Object Storage (COS). An all-legacy tech solution was easily outperformed by new, software-based data movement technology. For instance, the EMEA bank was able to revolutionize handling of image copy archive data, transitioning from an ingestion throughput of 70MB/s to 908MB/s - a 17x increase.

These results, not just minor improvements but in many cases an order of magnitude improvement, clearly argue for breaking free from pocket protector-era practices and bringing mainframe data management and data backup into alignment with modern capabilities.

The payback is clearly there and achieving these kinds of results is largely a matter of overcoming institutional inertia.

##

ABOUT THE AUTHOR

Gil Peleg 

Gil Peleg, CEO of Model9, has more than two decades of hands-on experience in mainframe system programming and data management, as well as a deep understanding of methods of operation, components, and diagnostic tools. He is a co-author of eight IBM Redbooks on z/OS Implementation. He holds a B.Sc. in Computer Science and Mathematics. www.model9.io

Published Monday, August 09, 2021 11:38 AM by David Marshall
Filed under:
Comments
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
Calendar
<August 2021>
SuMoTuWeThFrSa
25262728293031
1234567
891011121314
15161718192021
22232425262728
2930311234