Although tape has been used in the mainframe environment for over half a century, it is still a cost-effective medium for storing and retrieving data. Today’s tape media, drives and automated libraries offer the capacity, reliability, and performance necessary for storing and protecting the ever increasing amount of data generated by business in our information centric world.
The very first tape storage product offered commercially was the IBM 726 Magnetic Tape Recorder that was announced in 1952. Prior to this time, data was stored on punched cards -- lots of punched cards. Each card had 80 columns of data with one byte of data per column. The cards could be read at approximately 100 cards per minute. This provided a read data rate of approximately 133 characters per second.
The IBM 726 was able to transfer data at approximately 7,500 characters per second, or 56 times faster than punch cards. With this technological breakthrough, the first commercially viable large system storage device was born.
Tape predates disk as the first commercial disk drive, the IBM 350 RAMAC was not introduced until four years later in 1956.
Both disk and tape have improved and evolved throughout the years. However, since tape was first available, it has served the industry well as the principle storage device for large files of sequential data. This is true even today as large data files continue to be written directly to tape.
Tape is also used today as the principal means of storing backup data for both on-site and off-site recovery purposes.
The first tape media was an open reel-to-reel system, using a linear recording method. In the 70s and early 80s, the half-inch open-reel systems evolved to use a closed, single-reel cartridge. The size of this tape cartridge was a 5 ˝-inch square by about 1-inch high. This continues to be the size of most tape cartridges used in the large systems environments today.
Many improvements in the tape media and capacity have been gained over the years: When the first cartridge tapes were announced in 1984 (along with IBM 3480 tape drives), they had a capacity of just 200MB. By 1991, tape capacity had grown again with the introduction of 3490-E drives and new cartridges. The native capacity of the “E” or extended-length cartridges was 800MB. With the addition of IDRC compression (a hardware feature on the drives), the cartridges could hold up to 2.4GB of data – the highest capacity available at the time.
Although the 5 ˝-inch cartridge has become standard, the media contained within varies depending upon the tape hardware vendor. By 2005, StorageTek (Sun/STK) offered a 200GB native capacity cartridge, while IBM offered a 300GB native cartridge. (Unfortunately, neither the drives nor the media are interchangeable between vendors.) Assuming a 3:1 compression ratio, nearly 1TB of user data can fit onto a single tape cartridge.
Automation is one of the most important enhancements that has allowed tape to continue being a viable storage medium through the 1990’s. Without automation, it is nearly impossible to retrieve data in a timely fashion. Consider an environment with many thousands of tape volumes and a staff of tape operators who must locate the correct tape and mount it onto the tape drive. It was not unusual for some installations to have tape mounts wait for an hour or longer for the correct tape to be located!
Automation solved this problem: StorageTek was first to market with a viable tape robot when they delivered the first tape silos (Automated Cartridge Store or ACS) in 1987. The walls of the silo contained the tape drives and storage for perhaps 5000-6000 tape cartridges. The robotic arm would swing around in the middle of the silo, locate the correct tape and mount it on the drive. In support of even larger configurations, multiple silos could be bolted together allowing tape cartridges to be passed from one silo to another.
Of course, as the number of tape volumes continued to grow, it was not always cost effective to continue to increase the number of automated libraries. Thus, seldom used or long archive volumes were often removed from the automated libraries and placed back into the manual tape racks.
The virtualization of tape was the next important step that has allowed tape to continue as an effective storage solution today.
When considering just the physical media for a moment, it becomes apparent that one of the challenges is to utilize the capacity of the physical media to the point where it can continue to be a cost-effective storage option. Backup processes tend to maximize the usage of the storage media, but the direct use of tape – as it was first introduced into the “large systems” environment – is subject to a variety of processing irregularities.
Consider a data file that on some days contains thousands or millions of records, yet on other days, just a few. Or perhaps, a monthly transaction log that begins the month with just a few records, but increases exponentially throughout the month.
How can these types of files make efficient use of the physical media?
The answer, of course, is the virtualization of tape.
Virtualization of tape resolved two specific issues that had dogged the use of tape since its inception. These two issues can be summarized as:
The Virtual Tape Library (VTL) consists of an automated tape library, a few high capacity tape drives and a large disk buffer. Code running within the VTL allows the VTL to appear to the application processing system as many tape drives (256-512 or more depending on the model and installation choices) and many thousands of logical volumes.
With several hundred logical tape drives defined to the application system, tasks no longer need to wait in order to gain access to one of the limited number of physical tape drives. Instead, the application can immediately begin processing by accessing the disk buffer. If the task is writing an output tape, a logical volume is assigned within the disk buffer and the task proceeds as if it were writing to a physical tape.
Processing is slightly different if the task is going to read a logical volume that had been created earlier. First, the disk buffer is checked to see if the logical volume still resides on disk. If the logical volume is found, it is assigned to the logical tape drive and processing proceeds. If the logical volume was not found in the disk buffer, the VTL automatically locates the data on the correct high-capacity physical tape cartridge, copies the virtual volume to the disk buffer and then processing proceeds.
All of this processing is quick, automatic and requires no operator intervention.
Periodically, logical volumes residing in the disk buffer that have not been copied to the physical tape media are copied. Many logical volumes are copied to a single high-capacity volume in order to maximize the usage of the physical media.
This document was printed from http://recoveryspecialties.com/