As discussed in our last blog, tape, when correctly integrated into the disk architecture, does not negatively impact the performance of the backup infrastructure. In fact in some cases tape improves it while also providing a more efficient, more reliable long-term storage capability. Another advantage that organizations will realize when integrating tape into the backup architecture, is a significant price reduction of the backup infrastructure.
Since the majority of restore requests are from the most recent set of backups, the disk backup staging area only needs to be large enough to store the most recent backups. Most disk backup systems, however, are actually sized to store years, if not decades, of backups. The organization either ends up buying multiple disk storage systems or is constantly forklift upgrading those systems or investing in a more complicated and expensive scale-out backup architecture. Integrating tape into the backup architecture enables the organization to eliminate all of this complexity and much of the cost. A simple scale-up NAS or low node count object storage system will more than suffice.
Organizations can use these savings to invest in a high-performance ingest tier because restore requests from the most recent backup sets also tend to be the most time sensitive. The need for a higher performance is especially evident for organizations that are leveraging the in-place / instant recovery capabilities of most software applications. A small all-flash system with basic features is ideal for this use case.
The Tape Advantages
In the previous blog we demonstrated how tape could integrate into the existing backup architecture. Now let’s look at the financial impact. First, tape does have an upfront cost. The organization must buy the tape library and the drives to go in that library. Depending on the requirements for making a tape copy however, the number of drives may be limited to three or four. After making that investment, the savings begin and multiply.
What is a Library?
A tape library is a chassis that contains a number of slots (shelves) that hold tape cartridges. There is at least one robotic arm, and in some larger systems two, which will automatically pick a tape cartridge when the software requests it and load that cartridge into one of the available tape drives. When the software finishes with that cartridge, it will return the tape cartridge to a slot in the library.
What is a Tape Cartridge?
A tape cartridge is a storage device, which contains a spool of magnetic tape and is normally used in an automated tape library to store many different kinds of data. There are primarily two formats in use today LTO and IBM Enterprise Tape.
Adding additional capacity simply requires buying more tape cartridges. Typically, 12TB plus of capacity is available for less than $150. In most cases, the capacity of the tape media doubles since most drives have built in compression and greater than 2:1 compression ratios are not uncommon. Tape media today is available pre-labeled with a barcode. The software and robot use that barcode to request the right media and to track that media’s use.
The Savings of Integrating Tape
The first obvious savings is the cost of the media, especially when compression is factored in. Another savings is when a tape cartridge is not in use, which is its most normal state, since it requires no power. Idle data on hard disk drive systems use power continuously and the industry has long given up on spinning down hard disk drives.
In most cases, the organization will make two copies of the same data on two different pieces of tape media. One of these copies will go off-site and another will stay in the library. The off-site tape should be stored in a secure, temperature controlled facility and managed for easy retrieval. Most software applications that support tape also manage the tracking of off-site media and some vaulting locations can also provide tape media management services.
The cost of creating a second copy and storing it in a secure offsite facility is far less expensive than replicating a massive disk backup storage system to a second location, which requires a second system. It also requires keeping both systems powered on. Alternatively, some disk backup vendors may offer public cloud storage as an option. While the public cloud does have lower upfront costs, the recurring cost of storing data (especially data that doesn’t change) for years or decades will quickly surpass the upfront cost of the library. There are also egress fees, which can be substantial, to consider should it become necessary to retrieve any data from the cloud repositories.
It is also important to consider that most libraries remain in service for more than ten years, so the upfront cost pays off for a long time. The organization can expand tape capacity until the slots in the library are all used. Most libraries have some form of expansion that enables them to add slots. The organization has the option to choose to upgrade the library to keep more tape online or to off-site the tapes as they age. It is critical that the tapes are correctly transported and stored so that, if and when, they are needed in the future the organization can use them.
At some point in the life of a tape library, the organization will upgrade from one generation of drives to another to take advantage of increased capacity and performance. These are routine occurrences and most tape drives are backward compatible with previous generations.
The savings of implementing tape within the backup infrastructure are numerous; this blog only scratches the surface. While there is an upfront cost of investing in tape, and potentially a small amount of additional operational overhead, the savings and long-term security of integrating tape into the backup architecture are undeniable.