It seems no matter how often I clear my desk, eventually it gets filled with clutter. The more clutter the more time it takes to find anything. Eventually I take dedicated time and clear the desk of clutter, so it can become usable again. Managing Oracle/STK VSM back end tapes has the same challenge.

VTVs, or Virtual Tape Volumes are stored on MVCs, Multi-Volume Cartridges or stacked volumes. Eventually these stacked volumes become full and time must be taken to create usable free space. How do you know if the stacked volumes are used efficiently? Are too many not usable? Is too much of the space fragmented? When is migration going to stop because there is no more space?

Oracle/STK VSM software does not automatically manage the reclamation of fragmentation. Left to its own, most MVC pools would run out of space causing VTSSs to fill and tape processing to stop. This leads to buying more than are needed or needing to micromanage the tape pools.

Oracle/STK VSM has limited reporting for managing these pools and limited information about fragmentation. Reporting is based on static data, and trending of usage is missing even though it is a very important aspect of determining when more MVCs are truly needed.

Automate How You Monitor the Health of the MVC Storage Classes

IntelliMagic Vision analyzes  MVC XML files and automatically produces graphs with the information necessary to manage the MVC pools.  A monthly review can be used to determine how the MVC usable space is being maintained.  If trending shows that corrective action should be taken, it can be accomplished before a problem occurs during the middle of the night or the weekend.

As shown in the following graphs, trending of data is very useful to show how over time changes to MVC usage affects availability of usable capacity. IntelliMagic Vision provides an automated way to check the health of the MVC Storage Classes and the continued availability of MVC space so that you can reduce unnecessary spend and achieve optimal throughput and performance.

The following is a set of best practices which can be used to efficiently manage the VSM storage classes.

Best Practice 1: Know the number of MVCs available

Number of Stacked Volumes
Figure 1: Number of Stacked Volumes

 

The MVC count should remain stable. The graph above shows the number of stacked volumes, any change to the count should be investigated such as why the stacked bars on the far left are slightly higher before it levels off. The bars on the right are much higher because new volumes were added and are reported in an “UNKNOWN” Storage Class until they are assigned to a Storage Class.

Best Practice 2: Know the growth rate for both number of VTVs and space used

Number of Active Virtual Tape Volumes
Figure 2: Number of Active Virtual Tape Volumes

 

The graph above shows the number of active Virtual Tape Volumes which can be a good indicator of growth. Small sized VTVs took little space in legacy formated MVCs, a megabyte was a megabyte of tape space. With the new partition formatted MVC however, each partition is either 9.3 GBs on T10KC media or 11.2 GBs on T10KD media. On a fragmented partition this 1 MB VTV can allocate the entire partition. It is no longer enough to simply know the number of VTVs or amount of data which needs to be stored.

Space Used by Current VTVs in GiB
Figure 3: Space Used by Current VTVs in GiB

 

The above chart is the space used by current VTVs in GiB.  This chart shows that MVC used space will increase over time, but some Storage Classes may increase faster than others as shown in the above graph. It is important to review each Storage Class individually.

MVCs are assigned to only one Storage Class; all Virtual Volumes on the MVC must belong to the same Storage Class. Even when all Virtual Tape Volumes have been deleted it is still assigned to that Storage Class. MVCs must be drained before a Virtual Volume from a different Storage Class can migrate to it.

Best Practice 3: Know the usable space available for Migration

Total Usable Free Space
Figure 4: Total Usable Free Space

 

Usable free space must be monitored. When a Storage Class’s usable free space reaches zero, VSM can no longer migrate to it without additional MVCs, even if free space exists in another Storage Class. In the above graph, usable free space is constantly decreasing.

Total usable free space only increases on the right when new MVCs are added, but free space within each Storage Class remains unchanged. VSM will assign these new MVCs when usable free space is needed.

Best Practice 4: Know if MVC space is disappearing one MVC at a time

Number of MVCs in each Storage Class with an Error Status
Figure 5: Number of MVCs in each Storage Class with an Error Status

 

VTSSs will not migrate to MVCs with error statuses. The above graph shows the number of MVCs in each Storage Class with an error status. These error statuses can include data checks, audit or drain failure, broken, warranty expired volumes and other problems. Some of these errors can be corrected, others require the tape to be replaced.

MVCs with the Lost status are not permanently in error since migration to these MVCs will periodically be attempted. The Logutil For_LostMVC command can be used to recover VTVs for those truly Lost.

Best Practice 5: Know how fragmentation can waste away at the amount of usable MVC space

Fragmentation Percentage
Figure 6: Fragmentation Percentage

 

Fragmentation is normal but must be managed or it can consume the MVC pool until all MVCs are unusable. The fragmentation percentage shown in the previous graph can be different for each Storage Class. VTV sizes, expiry dates, number of MVCs and reclaim parameters all influence the fragmentation of each MVC pool. Knowing the fragmentation percentage isn’t enough, intense reclaiming of Storage Classes with a small number of MVCs may not yield a high return.

Amount of Fragmented Data
Figure 7: Amount of Fragmented Data

 

The amount of fragmented data is an indicator of which Storage Classes to focus on first. Review Storage Classes which have a high amount of fragmented data. Dynamic reclaim with MOVEDATA=NONE will reclaim empty partitions without mounting the tape cartridge. And reclaim with MOVEDATA=MINIMAL will reclaim partitions with a very small amount of data to move but does require the MVC to be mounted.

IntelliMagic Vision provides an automated way to quickly check the health of the Oracle/STK VSM MVC Storage Classes.

This article's author

Merle Sadler
z/OS Virtual Tape Consultant
Read Merle's bio

Share this blog

Oracle VSM Performance Analysis

This paper addresses correlating the virtual tape hardware metrics with the SMF tape workload data to get a clear picture of how the virtual tape hardware is handling the tape workloads and the replication between virtual tape hardware boxes.

Related Resources

Blog

Best Practices for Monitoring Oracle/STK VSM Usable MVC Space

Automate how you monitor the health of the MVC Storage Classes by following these 5 best practices.

Read more
Blog

How to Avoid Application Infrastructure Performance Problems

"What are the top 5 million things you need to do today to avoid application infrastructure performance problems?"

Read more
Blog

Application Design Issues Cause Low Throughput to Virtual Tape

IntelliMagic Vision proactively identifies risks in your virtual tape environment and highlights potential issues before they have fully developed.

Read more

Go to Resources

Subscribe to our Newsletter

Subscribe to our newsletter and receive monthly updates about the latest industry news and high quality content, like webinars, blogs, white papers, and more.