CLOUD STORAGE AND ITS TYPES
In this article you will learn about the basics of cloud storage and various types of cloud storage. You will also get familiarize with the four main types of cloud storage. So let’s start with the basics of storage on cloud without any delay.
Basics of Storage on Cloud
Cloud storage is a model of computer data storage in which the digital data is stored in logical pools, said to be on “the cloud”. Certain storage must be attached to a compute node before the storage can be accessed, whereas other storage types can be directly accessed either through the public Internet or a dedicated private network connection. Cloud providers host, secure, manage, and maintain the cloud storage and associated infrastructure to ensure you have access to your data when you need it.
Cloud storage services allow you to scale your capacity as you need so you only pay for what you provision, usually on a ‘per gigabyte’ basis. The cost of storage will vary by type but in general, the faster the read / write speed of the storage, the higher the per gigabyte cost.
Cloud storage is available mainly in four types –
1. Direct Attached,
2. File Storage,
3. Block Storage and
4. Object Storage.
1. Direct Attached storage
sometimes referred to as ‘Local Storage’ Direct storage is storage which is presented directly to a cloud-based server and is effectively either within the host server chassis or within the same rack. This storage is fast and normally only used to store a server’s operating system, although it can have other use cases.
The main two reasons why direct attached storage is not so great for other uses besides to store the operating system is that it’s typically ‘Ephemeral’ — meaning that it only lasts as long at the compute resource it’s attached to — it cannot be shared with other nodes and while you can use RAID techniques, it’s not as resilient to failure as other types of storage.
2. File storage
File storage is typically presented to compute nodes as ‘NFS Storage’. NFS stands for Network File System and means that the storage is connected to compute nodes over a standard ethernet network. NFS-mounted storage is common-place but it tends to be slower than either direct-attached storage or block storage because the data travels over an ethernet network.
It also tends to be lower cost than either direct attached or block storage. One advantage of File Storage is that it can be mounted or used on multiple servers at once. File-based storage is a simple, straightforward approach to data storage and works well for organizing data in a hierarchical folder structure, that desktop users are familiar with.
3. Block storage
Block storage is presented to compute nodes using high-speed fibre connections, which means that read and write speeds are typically much faster and reliable than with file storage, making block storage suitable for use with databases and other applications where disk speed is important. You typically provision block storage in ‘volumes’, which can then be mounted onto a compute node, which it then effectively sees as another hard drive. Volumes can normally only be mounted onto one compute node at a time.
With both File and Block storage, you may also hear the term ‘IOPS’. IOPS stands for ‘Input/Output Operations Per Second’ and relates to the speed of the storage or to put it another way, how quickly data can be read from or written to the storage. We’ll cover this in a little more detail. Persistence is a term that is used when provisioning File or Block storage and relates to what happens to the storage once the compute node it is attached to is terminated. If the storage is set to ‘persist’ then it will not be deleted along with the compute node, meaning that it and its data are preserved and available to mount onto another compute node, though you will continue to pay for the storage. You can also, in some cases, set the storage so that it is automatically deleted with the compute node that it is mounted onto — in this case, as we know, it becomes Ephemeral Storage. Here, you will also stop paying for the storage but you will lose any data unless it is backed up somewhere. There are several ways to backup data in the cloud but one way to back up both File and Block storage is to take a Snapshot. As the term implies, this is a point in time image of the storage. Snapshots are usually fast to create (they don’t actually write any data, or rather they create metadata), don’t require downtime and subsequent snapshots record only changes to the data. They are great for returning storage to the way it was at a particular snapshot, though note, they cannot be used to recover individual files.
4. Object storage
The fourth kind of storage is Object storage. This is a different type of storage in so much as it’s not attached to a compute node, rather it is accessed via an API. Of all the storage types, Object Storage is by far the cheapest and also the slowest in terms of read and write speeds, but it is infinite in size to the end user.
Unlike File and Block storage where you provision a certain storage capacity and it fills up over time, with Object Storage you can keep adding files to it and it never fills up, you just pay for what you use. This makes Object Storage a fantastic repository for all sorts of unstructured data types, large and small, including documents, video, logs, backups, data from IoT, application binaries and virtual machine images.
Summary of the article :
· Cloud storage is available in four main types–Direct Attached, File, Block, and Object Storage. These storage types differ in how they can be accessed, the capacity they offer, how much they cost, the types of data they are best suited to store, and their read-write speed.
· Direct Attached (or Local) Storage is storage that is presented directly to a cloud-based server and is effectively either within the host server chassis or within the same rack.
· File Storage is typically presented to compute nodes as a Network File System (NFS), which means that the storage is connected to compute nodes over a standard ethernet network.
· Block Storage is presented to compute nodes using high-speed fiber connections, typically provisioned in volumes, which are mounted onto a compute node.
· Object Storage is accessed via an API and doesn’t need an underlying compute node.
