Automation: Keeping The Humans Away From The Robots By Larry Walsh 2150 words deck: Robotic tape systems have revolutionized the data center industry and dramatically improved information management. The simple fact is that automation is not enough, however, you need to better manage the data in an automated robot to obtain maximum efficiency. Is your automation project in the data center becoming an increasingly people-intensive operation? Many companies are discovering an unanticipated result: The installation of robots has not solved all the automation problems associated with cartridges. Manual intervention seems to be increasing and operations personnel are running ragged trying to keep the robot going. There is a new and unique challenge that automation and robots brings to the data center: How do you keep escalating personnel costs from diluting the cost/benefit justification for large-scale data center automation? The robot has revolutionized the data center industry and has improved information management in the data center dramatically. The simple fact is that automation is not enough, however, you need to better manage the data in an automated robot to obtain maximum efficiency and automation from Robotic Tape Systems. Does Automation Eliminate All The Human Intervention? In most cases studied, data centers that have invested in robots have found that the amount of human intervention needed to maintain these robots is much higher than first anticipated. One of the major reasons for this is the need to eject and enter cartridges to keep the optimum balance of open slots and scratch cartridges. As data sets are determined to be no longer needed in the robot, the cartridges must be ejected from the robot, filed back into an external slot and a new cartridge pulled from an external slot and entered into the robot. This can be a time consuming process, and it takes human intervention. Data Consolidation Tools The Key To Success By automating as many tape mounts as possible, thus reducing manual mounts, robots can achieve levels of efficiency that were previously unattainable. We will explore some key concepts, operational procedures and tools that can help you improve your automation efforts. The consolidation of tape data sets to achieve higher cartridge density has been made possible by data consolidation tools. Data centers need to understand the ways these new tools impact day-to-day operations in an automated environment, ways to fully exploit the technology and what new tasks are required. Typically, data centers that have invested in robots have a predetermined number of days that cartridge data sets need to be kept in the robot. This number is used in an effort to keep as many as possible, if not all, of the cartridge mounts inside a robot, and that is the key to fully exploiting the robot's capabilities. Data Consolidation Tools are essential to maintaining automation efficiency. Here are some real-world scenarios facing automated data centers today. Using Intelligent Data Set Consolidation Tools Example: Determining Minimum Active Data Set Levels Since there is a limited amount of space or cartridge slots in a robot, it is important to determine how much active data needs to be kept in the robot to reduce manual mounts. A data center ran an analysis using SMF data that determined if 30 days worth of cartridge data sets could be kept in the robot, 100 percent of all private (specific) cartridge mounts and 85 percent of all scratch cartridge mounts could be handled by the robot. Because scratch mounts can be handled with auto loaders in an outside drive, they are much easier to handle than private mounts outside the robot. This works well as long as the data center keeps 30 days worth of active cartridge data in the robot. However, in this case, because of increased business, miscalculations or other reasons, the 30 days active cartridge data level began to decrease. When the number of active days decreases, two things usually happen, and both are not good. First the number of ejects and cartridges requiring manual handling goes up; and second, the number of manual mounts goes up. For data centers that have all the cartridge drives in the robot, ejects increase dramatically. Intelligent Data Set Consolidation tools can reduce manual mounts and keep active data in the robot by taking less frequently accessed data volumes and stacking them consecutively inside the robot on existing cartridges or scratch cartridges, thus freeing up large quantities of cartridge space. Doing this enables a data center to keep many more active data sets in the robot and at the same time reduce the manual mounts and/or ejects that need to take place to keep a robot filled with active data. Example: Reducing Manual Mounts This data center's corporate objective was to keep 30 days worth of active data sets inside the robot. However, once the robot was installed and became active, the 30 days could not be maintained; to keep open slots and scratch cartridges at the level needed active days kept being reduced. This reduction in active days kept within the robot created an increase in manual mounts and a large increase in ejects and inserts into the robot. The data center decided to look at the benefits of stacking data sets inside the robot. A careful analysis of data sets inside the robot that had not been opened in 14 days, indicated that 1,800 data sets inside the robot could be stacked onto existing cartridges inside the robot. With 1,800 slots now freed, the data center was able to move back and maintain 30 days of active data inside the robot. As a result of using a consolidation product, this data center was able to keep all of its active data sets inside the robot, reducing manual mounts, and eliminating the eject and enter process that was taking place. The data center decided that data sets not used in the last 14 days were good candidates to be consolidated, and that data sets still needed in the robot that are less used could be consolidated onto other cartridges. By using a consolidation tool and taking these data sets and stacking them together in the automated robot, slots were freed so other data sets can also be stored inside the robot. The robot can now be viewed and managed as a pool of storage space rather than X amounts of slots. Example: Eliminating Overtime Expense For Backups Another data center had two robots and was able to keep private data sets inside the robots, but, because of the amount of scratches used, did all of their disk back-ups on Saturday nights. This required two operators for two shifts loading scratch cartridges back into auto loaders outside the robot. The data center had to find a way to get 1,600 additional scratch cartridges into the robot. If this could be accomplished, then the disk backup could be done inside the robot. In addition to saving 32 hours of overtime each week, the disk backup itself could be done in less time. A Data Consolidation Tool analysis of the data sets in the robots indicated that data sets not used within 15 days could be consolidated freeing up a total of 1,654 slots so that scratch cartridges could be added. This particular data center invested in the consolidation and eliminated unnecessary overtime while at the same time improved performance. There is an Alternative: Use a Data Consolidation Tool. With consolidation tools, data centers are identifying data sets no longer needed in the robot, and are stacking smaller data sets or volume moving larger data sets to scratch cartridges outside the robot. That is they are logically ejecting the data rather than physically ejecting the cartridge. Example: Logical Eject versus Physical Eject A data center had 1,000 data sets identified that needed to be ejected from a robot to get to the proper scratch cartridge level. When these data sets are physically ejected, a person or persons has to handle 2,000 cartridges: 1,000 that are ejected and have to be taken out of the robot and filed into an external slot, and 1,000 scratch cartridges that have to be pulled from outside slots and loaded into the robot. Using a data consolidation tool these data sets can be logically stacked from inside the robot to outside the robot, with the actual time involved and the number of cartridges handled significantly reduced. For instance, if only an average of 5 data sets will fit on a cartridge, then the number of cartridges handled will be reduced from 2,000 to 400. (There will only be 200 scratch cartridges used on outside drives, then 200 cartridges filed away. See Figure 1.) The last step of the consolidation will be setting a flag in the Tape Management Catalog so the cartridges that are inside the robot from which data has been copied will be treated as scratch cartridges the next time an update is run. Data consolidation makes all this possible! Key Questions Answered What happens if I stack two data sets on a tape that are both needed in the same job? While this should be a concern, there are consolidation tools today designed to ensure against 513 abends or other similar downtime hazards. Using historical SMF data, a Conflict file is created that contains historical SMF record type 14 and 15 information to keep track of data sets used in the same job. When data sets are selected for consolidation, this file is queried and data sets used in the same job will never be stacked on the same cartridge. These tools are uniquely designed to protect your processing time investment while significantly improving automation efficiency. What are the alternatives to data set consolidation? This question has been asked many times since data set consolidation came to the market a few years ago. Of course it would be much easier to stack all cartridge data sets beforehand, but that never really happens! It's a fact that not all cartridge data sets are good candidates to go to DASD. There is a percentage of tape data sets that are excellent candidates to be moved to DASD, thus reducing tape mounts. However, many data centers are realizing that there is still a major need to manage tape data sets and it becomes an even bigger issue when robots are involved. Just as important is the fact that the projected numbers of cartridge data sets that can be sent to DASD never equals the real numbers. In the "real world" there is, and will continue to be, a need to consolidate tape data sets. If our data center's goal is to fully implement tape mount management (TMM) won't this eliminate the need for a consolidation product? Not really, since data centers need to implement a strategy immediately to tackle the automation of cartridges while they work on getting a TMM solution in place. Many data centers that have fully implemented a TMM solution have found that using consolidation tools complements their operational strategy by addressing the many data sets the other products just will not handle. A consolidation tool is not a replacement or an alternative to TMM, but a tool which enhances that product to provide total data storage management. How does Data set Consolidation work? Using the information obtained in the Tape Management Catalog, (Blocksize and Record count) the amount of space a data set occupies on a tape is calculated. Using this information along with other data center dependent selection criteria, these data sets are consolidated and stacked on existing cartridges with similar attributes and/or expiration dates. The TMC and OS catalog are automatically updated once these data sets are consolidated or moved. Data Consolidation Tools Usage Is Increasing Automation and robots are increasingly becoming an integral part of data center operations. Industry sources and our own projections indicate that tape stacking products and services will achieve exponential gains during the next two years. We believe the current 15 percent market penetration will reach approximately 65 percent by the year 1997. (See Figure 2). In typical data centers 30 percent to 50 percent of the tapes are virtually empty. Before robots became prevalent there was a feeling in the industry "that it's only a $5.00 per cartridge cost, and we can live with that." That argument doesn't hold much credibility if you're out of floor space or are constantly buying cartridges. The more cartridges you have, the more human intervention is needed to maintain them. When you are adding thousands of cartridges a year the $5.00 media cost, racks, space, and personnel expenses add up to an awful lot of time and money quickly. Now with robots added into the data center equation, and with slot prices exceeding $30 each, the savings from using data consolidation tools is much more apparent. These tools can give you the greatest value from your automation investment, and reduce costly human intervention. Was this article of value to you? If so, please circle Reader Response Card No. Larry Walsh is a technical consultant in the Storage Management Group at Mobius Management Systems in Chicago. He has more than 20 years of data processing experience and has successfully implemented storage management solutions for many data centers in the Midwest. callout; The last step of the consolidation will be setting a flag in the Tape Management Catalog so the cartridges that are inside the robot from which data has been copied will be treated as scratch cartridges the next time an update is run. Data consolidation makes all this possible! callout: As a result of using a consolidation product, this data center was able to keep all of its active data sets inside the robot, reducing manual mounts, and eliminating the eject and enter process that was taking place.