The source and explanation of short-lived cloud storage

The word "short-lived" means it may disappear. It was temporary, even fleeting. So why should I store something in a medium that might disappear without any warning? When all my requirements are more processors and more memory, how can I buy more of this storage?

This is a comment article written by Pete Johnson, senior manager of the Profitbricks company Evangelism platform, which has a funny name, don't you think? The term "short-lived" means that it may disappear. It was temporary, even fleeting. So why should I store something in a medium that might disappear without any warning? When all my requirements are more processors and more memory, how can I buy more of this storage?

Sources and explanations

Short-lived storage exists only because the first generation of cloud providers provide a large number of application servers. The business model is simple. The cloud provider buys a physical server and sells as many virtual machines as possible on the basis of this physical server. Since virtual machines are linked to physical servers in this approach, the first generation of cloud providers set the same virtual machine specifications to make it easier for them to do the business themselves.

However, in the process, these providers cannot do anything to improve the redundancy of the disk on the physical server and therefore cannot guarantee its availability. Instead, they tell you not to trust it entirely. It may disappear. What we are told is that "the code around it will be replaced".

If I can't trust it, how can I buy more of this storage when my operating system needs 10GB of storage and when I need a larger virtual machine somewhere else?

Look at the one with the largest capacity. Who would pay to buy a 1.6TB-capacity cloud storage?

Google Computing engine and profitbricks bring the reason

One of the best features of Google's computing engine is its ephemeral storage methods. Google calls this method temporary storage. In many cases, this storage limit is 10GB temporary storage per machine. This storage capacity is sufficient to establish a base for operating system use. Google apparently did it on purpose.

Profitbricks This approach to a step forward, does not provide a temporary storage. Instead, the physical servers that have the processor kernel and memory installed in a resource isolation pool are separated from the disk array that provides the block storage. Good IOPS (input/output per second) is achieved by connecting the two by using a 80GB InfiniBand network per second. In profitbricks mode, all storage is similar to high availability redundant block storage.

What you really need is a block of storage

One of the things that new public cloud novices find difficult to understand at first is the difference between a short storage and a block store. The latter is a feature provided by every IaaS (infrastructure or service) vendor, all of which has some level of redundancy, where data should be really stored. The following examples are the redundancy and better results provided by some vendors.

Lessons to be learned

The implication of this article is simple: if you put something in excess of your basic operating system on a temporary storage basis, you will have a lot of risk. That data may disappear at any time. You can't depend on it. Unless you use your engineering expenses to add extra forms of redundancy, you cannot use it. The data you care about belongs to block storage: it has built-in redundancy and improved usability. This will ensure that the data you care about will be there when you need it.