The need to perform backups in databases is becoming increasingly crucial for businesses due to the large amount of information stored. Because of this large amount of information is no longer feasible to store the backup on a hard disk, it is necessary to choose other alternatives for backup storage. One of the most widely used alternatives to the hard disk storage is the use of tapes, which allows for storage of large amounts of information in a reliable, fast access and with an excellent cost/benefit ratio. Yet, few companies use this technology because they don’t know its benefits and do not find detailed information on this subject in books and databases courses.
In this article we will discuss the requirements, technologies, strategies and recommendations for the recording of a database backup on a tape. In addition to presenting details about the recommended hardware and software this article also indicate two hardware solutions for the storage of tape backup.
This first part focus on the requirements and properties of magnetic tapes used to store databases backups, and talks about some existing hardware solutions to work with tapes.
In the DBMS (Database Management Systems Data) currently available data storage is done directly on hard disk, because this media type provides several advantages such as speed, scalability and availability of information. In medium and large databases is common to find various hard disk storage technologies such as RAID.
RAID (Redundant Array of Inexpensive Disks) is a technique that involves replication of logical disk volumes on different physical disks, separated in real time to ensure continuous availability. A mirrored volume is a logical representation of full copies of separate volumes. There are several types of RAID (RAID1, RAID5, RAID10 etc.) Each providing different levels of availability and performance.
However, the backup of these databases is not usually written directly to disk because of the large amount of information stored in the database. Other restrictions on the recording disk backup include the redundancy of information and also the possible vulnerability of backup, since it would be stored with the data itself. Due to these factors, companies have turned to external storage backup database to tape, as this media type has several benefits.
Besides allowing the storage of large amounts of data, much larger and cheaper than the space provided by the server hard drives or an external storage, tape allows the backup to be physically separated from the server that contains the database. Add these features to the life of a tape and we have a media that has an adequate storage solution that has been used since the old days of when mainframes dominate the database world.
Compared to other media such as CDs, DVDs, flash drives, external hard drives, and other, the tape has the advantage in almost all aspects, due to its technical specifications to be presented in this article. The DVD and other optical media in particular have become popular for data storage and personal backups, yet they do not offer a fraction of the reliability, speed and storage capacity provided by the tapes. In practical terms, it is rare to find a database of medium to large size that use a backup solution based on a type of media other than tape.
The use of tape backups requires a hardware investment because you must purchase a device that makes recording and reading of data. Also, you need the cartridges that actually contain the tape. This is a requirement that often limits the use of these media, since it is necessary to invest in hardware and software for the backup something that is not always easy to justify to the board of directors or the manager. In these cases it is worth remembering that more investments in backup will indicate that the company will be prepared for situations that require reconstruction or recovery of the database.
Once the decision on the storage of the database backup in tape has been taken it is necessary to evaluate the various technologies required. First you must choose the tape technology that will be used. Currently the most common tape technologies are known by the acronyms DTL, SDLT and LTO. Here it is worth mentioning that a DBA should know the aspects of hardware and software related to the backup, as this type of technology is part of the daily routines for who is involved with the database.
The DLT (Digital Linear Tape) is a tape that was created in 1984 by the company DEC (Digital Equipment Corporation). There are several versions of the DLT tape, all of which use a half inch coil that allows the recording and reading of data in several bands through a magnetic process. The main features of this tape are the ability hardware compression, resistance to one million reads, or about 10,000 copy operations, and life expectancy that can reach 30,000 hours. In the most recent models, such as the DLT-IV, the capacity comes to 160 GB (Gigabytes) with a throughput of 10 MB/s (megabytes per second). Figure 1 shows a cartridge produced by HP (Hewlett-Packard) containing a DLT-IV tape.
Figure 1. Cartridge with a DLT IV tape produced by HP.
The SDLT (Super Digital Linear Tape) is an evolution of the DLT. The difference between a DLT and SDLT is that this new version has an optical system in the cartridge that holds the track record of the back and front of the cartridge in sync with the read-write heads. This type of system makes the recording tracks thinner, allowing storage of more data. Just as the DLT there are several models of SDLT tapes. One model is the SuperDLT II, which stores 300 GB with a transfer rate 36 MB/s. Figure 2 shows a tape cartridge that contains Super DLTtape I produced by Quantum.
Figure 2. A tape cartridge Super DLTtape I produced by Quantum.
The LTO (Linear Tape-Open) is an open storage standard that was developed by IBM, Seagate and other manufacturers in order to reduce the monopoly of the manufacturers of DLT and SDLT tapes, which was basically held by HP and Quantum. Because it is based on an open standard and not rely on a specific manufacturer most companies quickly adopted the LTO technology making it the default type of tape used nowadays.
An LTO tape is composed of magnetic layers that provide a high recording density, a high throughput, and large storage capacity. Besides having more tracks of recording, LTO tapes incorporate a memory chip that stores the positions of the data along with the track position on the tape, which greatly reduces the access time to files. The estimated time of life for LTO is 30 years and it has the ability to withstand 1,200,000 passes, which are approximately 15,000 copies.
As well as DLT and SDLT tapes, there were several generations of LTO. Table 1 presents the main characteristics of the generations LTO-1, LTO-2, LTO-3 and LTO-4. It may be noted from the characteristics of the generations that the amount of storage and speed of data access are increasing rapidly. There is an expectation that future models go beyond the barrier of 1 TB (terabyte) of space and that the rate of transfer of these new models is greater than 200 MB/s. Figure 3 shows a cartridge containing an IBM LTO-4 with native storage capacity (without compression) of 800 GB .
Table 1. Characteristics of four generations of LTO tapes.
Figure 3. A Cartridge with the Ultrium LTO-4 tape produced by IBM.
An important detail about the DLT, SDLT and LTO is that both the storage capacity and transfer rate obtained varies with the device that reads and makes the tape recordings. To better understand how a drive for reading and writing works Figure 4 presents the main components of a generic drive. Each tape has specific features to its own drive, but the parts shown in Figure 4 provide a basis for understanding how the process of reading and writing data from a magnetic tape.
Figure 4. Main components of a device for reading and recording DLT, SDLT and LTO tapes.
When a user inserts a cartridge into the drive this cartridge is positioned with the tape's tip facing the tape guides. Then the drive automatically exits the beginning of the tape cartridge. Next, this tape passes through the guides and the read-write head in position and take-up reel. As operations are done reading and writing the take-up reel drives a small motor that advances or rewinds the tape, until the band is positioned right on the head, which is responsible for performing the required operation though a magnetic process. At the end of use is necessary to rewind the media (as with the old VHS tape) before ejecting it.
The drives that read and write the tape can be found in the form of an external device or coupled to a server, with the latter configuration is being more used. Manufacturers of servers typically design a line of products specifically for storage that are composed of hardware, software, training, support, warranty, and SLA, leaving the client with only the responsibility to choose how they will backup the data and administer the server. For example this article will present two hardware solutions from different manufacturers, which allow the storage of tape backup.
Although this article presents only two solutions for manufacturers, the market has several choices that must be taken into account according to need and available resources to be invested in the backup solution. The intent of this article is not to compare the solutions favoring one or another manufacturer, but present different examples of available products used for backup.
SLA: SLA (Service Level Agreement) is an agreement between the field service provider and its customers. This agreement should make clear which services are being offered (specific services) and the level of each service (hours of operation, downtime, time of support, etc.). Usually this agreement is put in the form of a contract that must be signed when hiring the service. For the database, in particular, one can use an internal SLA, where the DBA is committed to give some sort of feedback to the requester. Note that this feedback does not mean necessarily solve the problem or repair, but the DBA is aware of the request.
The HP alternative
HP has a line of dedicated servers for tape backup. Here we will see the details of a server whose full name is HP StorageWorks 1/8 G2 Tape Autoloader. The front panel of this server can be seen in Figure 5.
Figure 5. Front panel of the HP StorageWorks 1/8 G2 Tape Autoloader Server.
This server is a Tape Autoloader, i.e. it allows the internal storage of up to eight cartridges and writes to LTO tapes in the order specified by the software. In addition, the format of this equipment is called the backplane, as their physical dimensions allow it to fit into 1U (one unit of space) of a hack, which is usually where the servers are stored with these dimensions. Despite having two entries for tapes, this server has only one LTO driver, whose generation can be chosen. According to the technical specifications, this machine has a bar code reader to help identify the cartridge, encrypt the data via software, and has buttons on its front face to eject, rewind the tape and move inside, and colored LEDs that indicate the possible states of the tapes (full, empty, in writing, etc.).
This server, following a trend of major manufacturers, allows configuration of the main components such as processor, memory, disk space, network and tape drive technology. As this product is based on the standard LTO, you can choose the generation to be used from the generation LTO-1 that support up to 1.6 TB of storage with data transfer rate of 115.2 GB/hour to the generation LTO-4, which supports up to 12.8 TB of storage with a rate of 576 GB/hour. These values are obtained from the sum of all tapes stored inside the server with compression enabled.
From the point of view of software HP recommends the HP Protector Express Single Server Edition with versions for Windows, Linux and NetWare. In addition to the complete administration of each cartridge stored the software still has compression algorithms, recovery and management options for the Internet. An important detail is that even with HP's proprietary software can be used other solutions such as the popular BrightStor ARCserve Backup provided by Computer Associates (CA) some time ago and now provided by IBM.
The DELL alternative
The manufacturer Dell also has a line of products and services specific to store the backup in tape. Besides having Autoloaders Dell also has specific provisions for recording the backup that must rely on a different server to operate. This is the case with the Dell PowerVault 114T, whose front panel that can be seen in Figure 6.
Figure 6. Front Panel of the device that read and write tapes named Dell PowerVault 114T.
This device looks like a tape recording server backplane, but it is a subsystem that needs a computer to be used. In other words, this device has no CPU, memory or hard disk. According to its specification, Dell recommends using this device in conjunction with its line of PowerEdge servers or PowerVault coupled to other servers to build a NAS (Network Attached Storage), an external storage solution that has been adopted by many companies. A NAS has the benefit of separating the servers used exclusively for the storage of data thus organizing and facilitating the management of all computing resources.
The Dell PowerVault 114T takes 2U on a hack and allows the use of various tape technologies (DAT, DLT, SDLT and LTO). Regardless of the technology chosen for the two drivers available on this machine, the Dell PowerVault 114T allows storage of only one cartridge at a time in every driver, unlike the HP server that allows the internal storage of up to eight cartridges. If the choice of the driver is the standard LTO-4 this device will supports up to 1.6 TB of storage with transfer rate up to 160 GB/hour. In addition, two tape drivers can be used in parallel, which speeds up the backup process.
To manage the backup with the Dell PowerVault 114T the manufacturer recommends the use of Veritas NetBackup and Veritas Backup Exec for Windows and Netware, respectively. In addition to this software DELL also certified the Yosemite Tapeware available on Windows, Netware and Red Hat distribution of Linux.
An important detail of the HP StorageWorks 1/8 G2 Tape Autoloader and the PowerVault 114T is that both allow the use of third-party cartridges, i.e. it is not compulsory to use cartridges from HP or DELL. However, one must be aware of some details, such as the support for WORM (Write Once Read Many), especially when using LTO tapes, as these details may adversely affect the storage capacity or the rate of data transfer.
WORM (Write Once Read Many) is an exclusive feature of LTO-3 tapes. The tapes that have this feature record data on tape only once and allow several reads of what was recorded. The objective of WORM is to ensure that only one recording is performed on the tape, because this need is required to create concrete evidence in legal proceedings. Cartridges that have the WORM functionality contain a chip inside that implements this functionality and are usually painted with a color that distinguishes them from regular LTO cartridges.
Performing a backup of a database is a crucial task for any environment. Due to the amount of information, redundancy and reliability companies are storing the backup on tapes as an alternative to hard disk usage.
This paper presented the requirements, technologies, strategies and recommendations for the recording of a backup database into a tape. In addition, the article also pointed out two different solutions of the reading device and tape recording that provide a complete solution for backup storage.
To facilitate the planning and implementation of a backup strategy, four different types of generic backup strategies were presented with suggestions for scenarios easily found in companies. The commands, actions, details and recommendations for each strategy were explained, highlighting that each environment has its own characteristics and it was suggested that each strategy should be adapted according to the needs and requirements of the backup.
The investment in the backup process is fundamental. Knowing how to address the resources to this process is one of the actions that sooner or later will pay off because this process deals with the storage of critical information for businesses and that should always be available and operational. To achieve this goal is necessary to concentrate efforts to obtain resources, strategies and tactics related to the backup, as this action represents one of the most important contingencies that a company can implement in the computing environment.
This first part discusses about some details regarded to the use of tapes as a solution to store database backups, and explains some details about the hardware solutions provided by two manufacturers. The details about the hardware and the software were briefly covered in order to provide an overview to the reader.
Details about the HP Storage Works 1/8 G2 Tape Autoloader:
Details about the Dell PowerVault 114T:
Information about DLT e SDLT tapes:
Information about LTO tapes:
The second part of the article will cover four backup strategies that can be employed with Microsoft SQL Server 2008 regarding the backup.