Magnetic tape - the old horse does not spoil the furrow
Every time when we meet the phrase data center, or the data center abbreviation (data center), our mind instantly pulls out a set of standard patterns from the cache, which would seem to be unequivocally associated with this characteristic representative of modern IT infrastructure. Spacious rooms, server racks are littered with splashes of multicolored LEDs, the rumble of power supplies rivaling even more noise from the exhaust that removes excess heat from the rooms, intertwined cable bundles of various diameters and colors, engineers cutting through important corridors between the walls between the walls, built of high-tech equipment. What to speak about the huge electricity bills, it would all seem so natural and without any alternative. I will not disappoint anyone, in general, the way it is, in 99% of cases.
At the end of 2014, Spectra Logic, one of the IBM divisions, announced the creation of the fifth generation of 3592 series tape cartridges (type D), model 1150. This development allows you to store up to 10 TB of data, and an impressive 360 ​​MB / s read speed from the media. Moreover, this cartridge has compatibility with all existing IBM tape libraries since the 90s. Data center and data storage media, how do these two concepts relate in the world where HDD and SSD dominate? At the beginning of the article, we remembered a classic data center, of which the absolute majority is indeed, but that small percentage of IT nodes, of which it has become an integral component, many of us have already forgotten, magnetic tape, remain.
')
Data centers in which the rate was made on a magnetic tape, as on the main data carrier, appeared of course not today. Back in the 1950s, at the dawn of the birth of the industrial exploitation of such carriers, they very quickly gained popularity, and there were quite objective reasons for that. The first, and most important, is the ratio of the volume of data to be placed and the dimensions of the carrier. By the way, the first tape libraries contained no more than 2.5 MB of data, but the production was much simpler and cheaper than the first HDD solutions. Also, this technology allowed the use of a rather wide range of materials, which, with an adequate price, could become for the binary-coded data a safe haven for decades. As a matter of fact, for decades it was, huge reels with megabytes of data were the same characteristic features of data centers of the 1960s and 70s, which in the 21st century were HDD and SSD drives.
Let's take a closer look, so that tape drives, such as dinosaurs from the creation of the digital world, not only survived to our century, but also remained in demand for consumers of the IT infrastructure market. As at the very beginning of the use of tape magnetic drives, at the moment they have one of the best price-volume-compactness-durability. But one should not forget the fact that, besides the obvious advantages, the mentioned technology carries with it certain difficulties that significantly limit the range of use of tape media. Obviously, the cartridge recently introduced by Spectra Logic, is the brainchild of our time and with its predecessors from the 50s, has a similarity based only on the very principles of technology organization.
The combination of the “tape library” that came out of it is something more than a cabinet with systematized cassette blocks stored there. Now the market has a number of manufacturers of such IT equipment, the largest of them are IBM, Dell, HP, Compaq, Fujitsu - the real flagship of modern times in the high-tech industry. As a result, the complexity of the products and the ability to integrate them into the most advanced systems have reached real heights. Modern automated libraries completely eliminate human intervention in their activities, the speed of processing cartridges, thanks to the progress of mechanical design elements and software, has become acceptable for the organization based on libraries of autonomous backup services. Also, most libraries, say a manufacturer like HP, can be equipped with an external SCSI or Fiber Channel interface, and this in turn allows you to receive and send substantial amounts of data. High-speed data transfer interfaces have long been a necessity for this kind of solution. Prior to the recent presentation of the Spectra Logic 1150 tape cartridge, with a 360 MB / s read speed, cartridges with a maximum speed of around 250 MB / s were widely used, which by itself was not critical, for transferring large data arrays, as well as the reason for the use of multiple reading devices in automated libraries.
As for the minuses of tape storage systems, they are highly predictable. The central problem is the limitations of multi-user access to information hosted in libraries. This is especially aggravated by the fact that another section of the library can easily accommodate 600 TB in itself. Also, a very limiting factor is the sequential recording of information on cassette carriers, which naturally excludes the possibility of obtaining instant access to several data arrays.
Based on the strengths and weaknesses of the technical implementation of systems that are created on the basis of tape drives, they, in turn, have become an ideal repository for backups. They also found their application in research institutes, became an integral companion of "super computers" that generate petabytes of data. The explosion of the growth of high-definition television also played an important role in the demand for the “magnetic tape”, thousands of hours of capacious video created by multimedia companies left no alternative to cassette cartridges. Companies that own cloud services are also among the customers of this technology. The impressive growth of the “clouds” put the company's data in front of a reality to operate with huge data arrays within the boundaries of complex systems that are based on a multitude of clusters, and this in turn makes it possible to implement the cheapest data carrier at the moment - magnetic tape. According to various estimates, 1 GB of high-volume, tape media costs the buyer only 4 cents, against 10 cents for similar HDD media. When it comes to energy consumption, the usual HHD and SSD are simply not competitors to tape. “Cold storage” of data, and of course the lack of high heat generation gives the right to be called automated libraries the most energy-efficient integrated solutions for long-term data placement.
The fact that automated libraries were able to operate within the boundaries of a single operating system made it possible to very easily scale such storage, making it possible to combine dozens of disparate devices into a single structure.
CERN network engineer, Mike Collin (Micke Collin), speaks about the embedded node, mainly on tape technology: “In our farm there is no adequate tape replacement. As a result of our work, on the basis of a larger hadron collider, studying the structure of matter, pushing protons and ions at enormous speeds, a huge amount of data is generated. Not only do the experiments themselves imply the registration of all the information on the collision of millions of particles per second, moreover, we have to keep all this information in our data centers for years. From the very beginning of the organization of work, there was a task to minimize the loss of resources for the storage and processing of petabytes of data obtained as a result of multi-million expenses. And in this the magnetic tape became a faithful assistant. ” Focusing on the characteristic features of tape cartridges, Mike noted: “The uniqueness of the tape lies in the fact that in the event of a carrier failure, it has much greater potential for recovery than its competitors. There were cases when, for various reasons, the cartridges failed and we lost pieces of tape, but after removing the torn section and gluing the remaining ends, we had access to the absolute majority of information on the cartridge, while at the same time merging with malfunctions of SSD and HDD, there were precedents for the loss of the entire data array located on the disk ".
Naturally, the technology implies the storage of information on a magnetic tape is not a completely independent, self-sufficient data carrier in modern data centers. At the same time, it is an excellent complementary element of the network infrastructure, making it possible to save substantial funds both at primary costs and during operation. The development of this technology also speaks of its relevance in the IT industry market, making it even more interesting for owners of large organizations that they have to interact with particularly large amounts of information in their activities. Having begun its formation from the mid 30s of the XX century and now this method of data storage looks absolutely relevant, besides it has all the makings of a further successful existence.
At the end of 2014, Spectra Logic, one of the IBM divisions, announced the creation of the fifth generation of 3592 series tape cartridges (type D), model 1150. This development allows you to store up to 10 TB of data, and an impressive 360 ​​MB / s read speed from the media. Moreover, this cartridge has compatibility with all existing IBM tape libraries since the 90s. Data center and data storage media, how do these two concepts relate in the world where HDD and SSD dominate? At the beginning of the article, we remembered a classic data center, of which the absolute majority is indeed, but that small percentage of IT nodes, of which it has become an integral component, many of us have already forgotten, magnetic tape, remain.
')
Technology
Data centers in which the rate was made on a magnetic tape, as on the main data carrier, appeared of course not today. Back in the 1950s, at the dawn of the birth of the industrial exploitation of such carriers, they very quickly gained popularity, and there were quite objective reasons for that. The first, and most important, is the ratio of the volume of data to be placed and the dimensions of the carrier. By the way, the first tape libraries contained no more than 2.5 MB of data, but the production was much simpler and cheaper than the first HDD solutions. Also, this technology allowed the use of a rather wide range of materials, which, with an adequate price, could become for the binary-coded data a safe haven for decades. As a matter of fact, for decades it was, huge reels with megabytes of data were the same characteristic features of data centers of the 1960s and 70s, which in the 21st century were HDD and SSD drives.
Let's take a closer look, so that tape drives, such as dinosaurs from the creation of the digital world, not only survived to our century, but also remained in demand for consumers of the IT infrastructure market. As at the very beginning of the use of tape magnetic drives, at the moment they have one of the best price-volume-compactness-durability. But one should not forget the fact that, besides the obvious advantages, the mentioned technology carries with it certain difficulties that significantly limit the range of use of tape media. Obviously, the cartridge recently introduced by Spectra Logic, is the brainchild of our time and with its predecessors from the 50s, has a similarity based only on the very principles of technology organization.
Tape libraries
The combination of the “tape library” that came out of it is something more than a cabinet with systematized cassette blocks stored there. Now the market has a number of manufacturers of such IT equipment, the largest of them are IBM, Dell, HP, Compaq, Fujitsu - the real flagship of modern times in the high-tech industry. As a result, the complexity of the products and the ability to integrate them into the most advanced systems have reached real heights. Modern automated libraries completely eliminate human intervention in their activities, the speed of processing cartridges, thanks to the progress of mechanical design elements and software, has become acceptable for the organization based on libraries of autonomous backup services. Also, most libraries, say a manufacturer like HP, can be equipped with an external SCSI or Fiber Channel interface, and this in turn allows you to receive and send substantial amounts of data. High-speed data transfer interfaces have long been a necessity for this kind of solution. Prior to the recent presentation of the Spectra Logic 1150 tape cartridge, with a 360 MB / s read speed, cartridges with a maximum speed of around 250 MB / s were widely used, which by itself was not critical, for transferring large data arrays, as well as the reason for the use of multiple reading devices in automated libraries.
As for the minuses of tape storage systems, they are highly predictable. The central problem is the limitations of multi-user access to information hosted in libraries. This is especially aggravated by the fact that another section of the library can easily accommodate 600 TB in itself. Also, a very limiting factor is the sequential recording of information on cassette carriers, which naturally excludes the possibility of obtaining instant access to several data arrays.
Based on the strengths and weaknesses of the technical implementation of systems that are created on the basis of tape drives, they, in turn, have become an ideal repository for backups. They also found their application in research institutes, became an integral companion of "super computers" that generate petabytes of data. The explosion of the growth of high-definition television also played an important role in the demand for the “magnetic tape”, thousands of hours of capacious video created by multimedia companies left no alternative to cassette cartridges. Companies that own cloud services are also among the customers of this technology. The impressive growth of the “clouds” put the company's data in front of a reality to operate with huge data arrays within the boundaries of complex systems that are based on a multitude of clusters, and this in turn makes it possible to implement the cheapest data carrier at the moment - magnetic tape. According to various estimates, 1 GB of high-volume, tape media costs the buyer only 4 cents, against 10 cents for similar HDD media. When it comes to energy consumption, the usual HHD and SSD are simply not competitors to tape. “Cold storage” of data, and of course the lack of high heat generation gives the right to be called automated libraries the most energy-efficient integrated solutions for long-term data placement.
The fact that automated libraries were able to operate within the boundaries of a single operating system made it possible to very easily scale such storage, making it possible to combine dozens of disparate devices into a single structure.
CERN network engineer, Mike Collin (Micke Collin), speaks about the embedded node, mainly on tape technology: “In our farm there is no adequate tape replacement. As a result of our work, on the basis of a larger hadron collider, studying the structure of matter, pushing protons and ions at enormous speeds, a huge amount of data is generated. Not only do the experiments themselves imply the registration of all the information on the collision of millions of particles per second, moreover, we have to keep all this information in our data centers for years. From the very beginning of the organization of work, there was a task to minimize the loss of resources for the storage and processing of petabytes of data obtained as a result of multi-million expenses. And in this the magnetic tape became a faithful assistant. ” Focusing on the characteristic features of tape cartridges, Mike noted: “The uniqueness of the tape lies in the fact that in the event of a carrier failure, it has much greater potential for recovery than its competitors. There were cases when, for various reasons, the cartridges failed and we lost pieces of tape, but after removing the torn section and gluing the remaining ends, we had access to the absolute majority of information on the cartridge, while at the same time merging with malfunctions of SSD and HDD, there were precedents for the loss of the entire data array located on the disk ".
Naturally, the technology implies the storage of information on a magnetic tape is not a completely independent, self-sufficient data carrier in modern data centers. At the same time, it is an excellent complementary element of the network infrastructure, making it possible to save substantial funds both at primary costs and during operation. The development of this technology also speaks of its relevance in the IT industry market, making it even more interesting for owners of large organizations that they have to interact with particularly large amounts of information in their activities. Having begun its formation from the mid 30s of the XX century and now this method of data storage looks absolutely relevant, besides it has all the makings of a further successful existence.
Source: https://habr.com/ru/post/253025/
All Articles