Skip to main content

How Your VTL Can Help Save Money by Becoming a Cloud Object Store

The transparent cloud tiering and DS8000 Object Store features provide a valued cost savings solution to mainframe storage customers.

Two green dollar signs in two separate dark blue circles with green lines indicating migration and movement.

Today there are a lot of choices when it comes to storing your mainframe data. The need to store more data, for a longer period of time and in a cost-effective manner has been a focus area for decades. That data is typically moved off of primary storage to a more cost-efficient, secondary storage system like the TS7700 Virtual Tape Library. As seen in Figure 1 below, the data must pass back through the mainframe before it is sent to an offline storage system.

Picture1.pngFigure 1: Traditional tape processing

But what about the cost to process and manage the data going to that offline storage system? 
 
Studies have shown that for small data sets, about 33% of the overall CPU consumed for data management is for the actual data movement in between the two storage tiers. But for data sets that are hundreds of megabytes or larger, the CPU consumed for the actual data movement can account for 97% or more of the overall data management CPU. Eliminating this workload can open up the availability of more processing power for critical application work. 
 
The IBM Systems Storage group at the IBM Tucson Lab started working on an innovative back-end data management solution and, in 2017, z/OS DFSMS and DS8000 disk system introduced transparent cloud tiering. By having DFSMS and DS8000 transparent cloud tiering manage a portion of the back-end storage workload, z/OS can reduce the MIPS it uses for back-end storage and reserve it for more main application tasks. The targeted data for the transparent cloud tiering offering is DFSMShsm data.
 
This transparent cloud tier solution allows DFSMShsm to move targeted data, in the form of objects, from a DS8000 primary disk storage system directly to a back-end cloud storage device without the need for z/OS to select a tape, allocate drives, re-block data and stream the data to the back-end device itself. When DFSMShsm initiates the migration command using this solution, DS8000 reads the data from its storage as it does today but sends the data to the cloud storage device that is targeted instead of sending it back to z/OS for processing.
 
This method can significantly reduce the amount of CPU resources normally used on a z/OS mainframe and also allows for parallel access to the data, eliminates the need for recycles and eliminates the constraints of 16K block size limit that exist with tape volume processing today. 

No Need to Purchase Additional Hardware 

I know what you're thinking: “But, now I need to purchase another piece of hardware, a cloud storage device, to store this object data!" What if I told you no additional hardware is required?
 
In November 2019, TS7700 development delivered a DS8000 Object Store solution as part of their microcode release level 8.50.x.x. This allows the same TS7700, which today store this DFSMShsm data in the form of a logical volume, the ability to accept this data from a DS8000 in the form of objects.
 
As seen in Figure 2, data will move directly from DS8000 primary disk storage to a TS7700 virtualization engine in the form of objects without the need to pass through the mainframe.

Picture2.png
Figure 2: TS7700 as an Object Store

The TS7700 Object Store feature utilizes the existing Ethernet ports and disk  cache system to receive and store objects. Object data will coexist with logical volume data but is separated using the existing Logical Cache Partitioning function.
 
It also utilizes existing Ethernet ports located in each CEC of the DS8000.  The DS8000 can target one or two TS7700s in the same grid for synchronous data redundancy. 
 
When z/OS is configured with a Cloud Network Connection construct that points to a TS7700,  and when a migration is triggered for Migration Level Cloud (MLC) by z/OS Management Class, DS8000 will route that object data to the TS7700. 
 
The TS7700 will manage the object data under an object cache partition. Additionally, the TS7700 further separates the object data by the Cloud Connection Name, which originates from the z/OS Cloud Network Connection construct. This means more than one DS8000 can target a single TS7700 and the TS7700 will manage the data separately although within the same object cache partition. 
 
Utilizing the same hardware to store this same data eliminates the need to purchase and maintain another piece of hardware on the data center floor. It allows all data to be stored in one secure place. 
 
The transparent cloud tiering and DS8000 Object Store features exploit the natural synergy between z/OS, DS8000 and TS7700 IBM products and provide a valued cost savings solution to mainframe storage customers.
 
For more information see IBM Redbooks publication: IBM TS7700 Series DS8000 Object Store User's Guide Version 1.0
IBM Systems Webinar Icon

View upcoming and on-demand (IBM Z, IBM i, AIX, Power Systems) webinars.
Register now →