What is commodity hardware? Even if this technology is common, not everyone has a clear idea of that. That’s why we can use this article to make some clarity.
What is commodity hardware?
To put it in a way that anyone could understand, commodity hardware means cheap servers.
Commodity hardware means cheap servers.
This doesn’t mean scrap servers, not necessarily at least. Instead, it indicates servers that are affordable and easy to obtain, even in large quantities. The modern company needs computing power to process and store information. Of course, you can get that from a battery of servers. The thing is, any company needs that. Imagine a company in the ’60s, when all the information was on paper. At that time, companies heavily relied on paper to store information. Yet, no one thought that paper was a critical part of the business model. Yet, everyone used it. Paper was just there, its usage was taken for granted. In other words, the paper was a commodity.
Fast forward 60 years, things have changed, yet principles stay the same. At the end of the day, businesses are hungry for information, just like they have been for the last century. However, they now rely on modern technologies instead of paper. Now like then, what’s important is the information, not the way you process it. Thus, now the hardware is the commodity.
Focus on the application, not on the hardware
The focus is on the information, on the data, and on the applications that transform raw data into meaningful information. Applications may be complex pieces of software, but in the end, they need to run on some sort of hardware. Yet, they don’t care about it, they are in the mood of “just give me some hardware”. Since this is the mood, hardware must be easy to get. This ease translates mainly into two requirements: low prices and low lead time. You don’t want to pay a lot for hardware, and you want to have it as soon as possible. So, what is commodity hardware? It is the hardware you can get quick and cheap.
The era of supercomputers is over
Big Data is another buzzword of modern times. With all this digital information, any company has a lot of data to process. Big data means big computers to process them, right? Well, not quite. Thanks to new development paradigms, the industry is moving away from supercomputers. Rather than having one big system processing everything, we now want to have many servers, each processing a small chunk of that. That’s where commodity hardware comes into the picture.
Now, modern applications prefer parallelism. This means you can have lots of not-so-powerful servers instead of a single big one. If one server fails, you lose only one tiny part of your processing power. In the end, you will have better efficiencies and more availability. In case you need more power, you can simply add a server: the application is already parallel, and you don’t need to rethink your entire solution.
Commodity Hardware in Hadoop
Maybe, you asked yourself “what is commodity hardware?” after reading about Hadoop. Hadoop is an open-source solution by the Apache foundation that helps you achieve parallelism. It is one of the leading projects in the Big Data world and has driven the industry in its early stages.
![Hadoop is a leading technology in big data and its definition of what is commodity hardware is now accepted in the industry](https://accelerates.it/wp-content/uploads/2018/10/664px-Hadoop_logo.svg_.png)
![Hadoop is a leading technology in big data and its definition of what is commodity hardware is now accepted in the industry](https://accelerates.it/wp-content/uploads/2018/10/664px-Hadoop_logo.svg_.png)
The concept behind Hadoop is simple: you have several servers (the commodity hardware) and you distribute the load among them. This is possible thanks to Hadoop MapReduce, a special feature of this solution. Hadoop is installed on all the severs, and it then distributes the data among them. In the end, each server will have a piece of data, but no server will have everything. However, the same piece of data will be duplicated on two servers to protect against faults. Now that each server has its piece of data, you can process the data. When you do that, Hadoop tells each server to process its own data and give back only the results. That’s parallelism, many servers running together at the same time toward the same goal.
In a nutshell (TL;DR)
In case this brief article was too long, here’s the takeaway. What is commodity hardware? Commodity hardware is server hardware you can get at affordable prices, fast and in large quantities. That’s what modern companies use, even tech giants like Google.
What do you think about commodity hardware? Do you use it at your job? Let me know your opinions in the comments!