The second in a series of articles on backup-related issues reveals end-user trends in tape and disk strategies.
By Thomas M. Coughlin and Farid J. Neema
This article is the second in a series exploring end-user trends in backup and archiving, as well as business continuance and disaster recovery. Our analysis is based on survey results from more than 200 North American IT managers, representing sites relatively evenly distributed among storage capacity ranges.
A similar survey was conducted in 2003. Each of the respondents filled out survey forms with more than 200 questions. All of the participants met the criteria of having backup and/or archiving responsibilities for IT operations that store a minimum of 500GB of raw disk storage.
Most of the respondents were from 10 industries: finance/banking, health, manufacturing, retail distribution, government, education, consulting, transportation, media/entertainment, and telecommunications. Sixty of the responding companies have more than 2 petabytes (PB) of disk capacity, mostly in the transportation, entertainment, and telecommunications industries. The majority of the respondents in the education and consulting industries have less than 5TB of storage capacity.
The survey results show that a large majority of corporate data is still saved on tape, using periodic incremental and full backups. A very small number of respondents said they intend to totally eliminate tape in favor of disk-based backup and recovery.
The existing notion that tape capacity is four to eight times the capacity of disk in a given site is a myth, except for a few very large sites. The overall mean in our survey population was a ratio of 2.3 (tape to disk capacity), with a median of 2.0. The ratio of tape to disk increases with the revenue tier of the company. A number of factors influence the tape-to-disk capacity ratio, including disk utilization rates and the amount of data replicated on disk.
Tape capacities continue to increase at almost all sites, but at a slower rate than in previous years. For the surveyed population, nearline tape capacity will grow from 782PB to 922PB over the next 12 months, an 18% annual growth rate. Disk capacity growth is in the 23% to 26% range.
The figure above shows that the majority of IT organizations use tape autoloaders and/or libraries or silos for backup. Tape autoloaders are used by almost 50% of the companies (compared to 15% in 2003), followed by tape libraries/silos at 45% (also 45% in 2003). Disk arrays are used for backup by 32% of the surveyed companies (up from 9% in 2003).
As shown in the figure below, DLT tape is the most popular backup media and is used in more than half of the surveyed sites. The use of LTO and AIT tape media, as well as disk arrays, has grown significantly over the last year, while the use of DDS and Travan tape has decreased since the 2003 survey.
As shown in the figure below, nearline tape capacity is used primarily for backup and snapshots, followed by archiving. Surprisingly, 27% of the tape capacity in this survey is used for primary data.
There is still a major concern among IT managers about the reliability of backup-and-restore operations. Close to half of the respondents in the survey experience an unsuccessful backup rate of more than 5%. The survey respondents ranked reliability and integrity at the top of the list of their concerns with backups. File corruption is the greatest cause of restoration reliability problems, followed by the inability to find files, software problems, and system-related problems.
Tape remains the media of choice for backup and archiving, and DLT is still the dominant tape format despite significant encroachment by LTO and, to a lesser degree, AIT. Meanwhile, disk-based backup systems continue to gain in popularity, and the reliability issues associated with tape backup need to be more effectively dealt with in order for tape to retain its popularity.
The full backup-and-archive report includes more than 170 figures summarizing site, industry, and revenue characteristics of the surveyed companies, as well as information on disk and tape backup and archive trends and perceptions.
(For more details about the report, check out www.tomcoughlin.com, Technical Papers section. A companion report based on the survey is also available covering business continuance and disaster recovery.)
Thomas M. Coughlin is president of Coughlin Associates (www.tomcoughlin.com), a data storage consulting firm. He contributes to reports from Peripheral Research and Peripheral Concepts and is a co-organizer of the Network Storage Conference and Storage Visions conference. Farid J. Neema is president of Peripheral Concepts (www.periconcepts.com) and chairman of the Network Storage Conference.