What type of Storage Media should I be using?
One of the questions we see most at Macrium Software is: what type of storage should I be using? This may seem like an easy question on the surface, however, the more you investigate this question, the more difficult it becomes. Every storage media has its advantages and drawbacks, by weighing these up, you can decide what type of storage is best for your needs. Today we will be looking into some of the more common storage media.
Backups and Archives
One thing to consider when deciding the storage media, you are going to use, is whether it is for archiving or backups. A backup is a copy of data (typically stored in an alternate location) that can be used to recover in the event of disaster. An archive can be seen as a collection of long-term records designed to archive data usually to cover a possible future requirement for historical data Typically, archives will be stored for longer than 5 years, whilst backups are updated over time and will not usually be kept for this long.
Hard Disk Drive (HDD)
Hard Disk Drives are everywhere. You will find these inside your servers, home computers, laptops and gaming consoles. But how do HDDs work and why are they still so popular? HDDs work by using two or more magnetic heads, attached to a moving arm, which reads the data from one or more rigid rotating disks coated with a magnetic material. These magnetic disks are very popular for several reasons. HDDs are relatively cheap when compared to other storage media, pairing this with their high capacity makes them perfect for storing large amounts of data when very high-speed random access is not required, e.g. performing backups and general home use. HDDs also have a long lifetime, making them an all-round reliable media. However, the moving parts will eventually lead to them degrading and can be damaged if the disk receives a shock.
Solid State Drive (SSD)
Solid State Drives are rapidly becoming the preferred local storage device. SSDs work by storing persistent data on micro-chips (Flash Storage). Unlike HDDs, SSDs do not have any moving parts and do not contain any physical disk. Since SSDs do not need to wait for a reading arm to physically access data, there are vast performance improvements over HDDs. However, this performance increase comes at a greater cost and lower amount of storage for your money. This makes SSDs ideal for running an operating system off and storing ‘hot data’ (smaller amounts of data, accessed more regularly, e.g. work documents). SSDs were first introduced with a SATA interface, this being the standard for magnetic disks. This enables an SSD to be a drop-in replacement for magnetic disks but being designed around a data stream from a small number of magnetic heads, it doesn’t fully expose an SSDs higher bandwidth and parallelism. To correct this, NVMe was developed to better match SSD’s performance characteristics. The physical layer is PCIe with either a U.2 (which to date has gained little traction) or more commonly M.2. Alternatively, an NVMe storage device can be integrated directly onto a PCIe expansion card.
There have been questions regarding the longevity of flash storage, specifically due to the charge carriers (electrons or holes) being lost (detrapping) through the tunnel oxide. It appears to be is due a misreading of the JEDEC spec that appears to specify surprisingly short retention times. However, this is only in the vanishingly unlikely circumstance where the data is written while the device is at an unlikely cold temperature and then the device is stored at an unlikely hot temperature. In normal usage conditions, the power off retention times of SSDs is measured in years.
Magnetic tape is usually made of a thin, magnetised coating on a plastic film. Magnetic tape is less common than it used to be, although, new magnetic tape drives are still in production. The hardware required to write to and read from magnetic tape is expensive. Once this hardware has been purchased, the magnetic tape drives themselves (usually in the form of cartridges) are cheap. These cartridges can also hold massive amounts of data (up to 15TB). However, due to the linear nature of tape, the drives must move to the part of the tape where the data is stored to access it. This means that the sequential I/O of tape drives are extremely fast, while the random I/O is very slow. All these factors make magnetic tape the perfect media for archiving data at a reasonable price.
Optical Media — Dye Based
A, now seldom-seen variation, of optical media is the magneto-optical drive. These devices contained a ferromagnetic material sealed inside a plastic casing. A laser is then used to read data from the magnetic disk based on the magneto-optic Kerr effect (an effect where light is reflected differently based on the magnetized state based on the surface of an object). Magneto Optical disks were hampered by a small maximum capacity (256MB to 9.2GB), making them inefficient for storing large amounts of data. The magneto-optical disks had a high longevity and were highly reliable, however, were superseded by cheaper optical media and flash memory devices.
A quickly growing storage media is Cloud Storage. This works by transferring your data over the internet to data centres that are owned and managed by the service provider you chose. Since data is being sent over the internet, this method of transfer is tied to your network speed and can often be slow. This also means that if you lose network connectivity, you also lose access to your files that are stored in the clouad. The plus side to this is that your data is safe from hardware failure. Redundant copies of your data will often be stored at that data centre (and potentially other data centres around the world), meaning that if one of the disks containing the data is damaged, no data is lost. Since the storage is managed by a service provider, you will often have to pay a subscription fee. This can add up over time, making cloud storage one of the more expensive storage options. This also means that your ability to store data in the cloud relies on the service provider, if they decide to shut the service down, then you will not be able to store your data with them. This makes it impossible to know the longevity of this storage medium. However, due to your storage being manager by a service provider, you have the option to select different products depending on your needs. An example of this is Amazon Glacier, which is designed for archiving ‘cold data’, whereas Amazon S3 is designed for data that is accessed more often. This is reflected in the prices where, Amazon Glacier is cheaper to store data per GB, but also incurs a retrieval fee.