The development of server RAM
In our blog special on the topic of ‘server RAM’, this time we look at the development of server RAM. How did it come about that special RAM modules were developed for servers? Why are conventional RAM modules not sufficient for mainframes? When were standards such as DDR2 or DDR3 and specialised modules such as Registered RAM and ECC RAM modules introduced? And what advantages do the specialised server RAM modules actually offer? All the answers can be found in this article.
The history of RAM
However, these calculating machines are only remotely reminiscent of today's high-performance computers - not only in terms of design and weight, but also in terms of the components used. In the 1950s and 1960s, hard disks with capacities of 5 megabytes dominated the market. Just to give you a rough idea: that would be (assuming approx. 4000 characters per A4 document) around 1250 A4 pages. A 1 TB hard disc, which is very common nowadays, has a digital storage capacity of around 250 million pages.
And it was the same with RAM: One of the first ‘types’ of working memory, widely used in the 50s and 60s, was magnetic core memory. You can imagine small rings of magnetic material (ferrite) strung on thin wires - depending on how the ring was magnetised, it could store either a 1 or a 0. In contrast to its predecessors, the storage tubes (e.g. ENIAC vacuum tube or Williams-Kilburn tube), this technology was very reliable and retained data even without electricity - but it was also slow and expensive.
If you would like to delve deeper into the development of office machines and computer parts, we also recommend our previous blog posts: Evolution of data storage - from 40,000 BC to the present day and 30 inventions that are now an integral part of our everyday office life.
The wild 70s: Revolution through DRAM
From PC to server - from PC RAM to server RAM
Back in the 1960s, mainframes were used by companies and universities. Although these computers were useful, they were not very powerful. You can think of it as a rather dusty library: The information is there, but it takes time and effort to look for it - but in return you had the books all to yourself.
With the introduction of the Internet in the 1990s, not only did the available content expand, but countless people suddenly wanted to access it: in other words, our library was bursting at the seams, not only because of the books available, but also because millions of visitors wanted to access it every day.
The transition to SDRAM with ECC came in the course of the 1990s. In contrast to DRAM (asynchronous architecture), ‘Synchronous DRAM’ works synchronously with the memory bus. The clock is therefore firmly synchronised, which means that memory operations became more efficient and predictable. This enabled higher speeds, parallel memory accesses and - thanks to ECC - significantly improved reliability. Translated with DeepL.com (free version)
What is server RAM?
Server RAM is a special form of RAM and refers to the main memory that is installed in servers or mainframes. A detailed explanation of RAM in general can be found in our article ‘RAM upgrade in the server’.
Compared to conventional PC RAM, server RAM is characterised by significantly higher capacities as well as greater reliability and stability. The modules are specially designed for long-term operation under high loads - for example in data centres or for business-critical applications. For this reason, they often support the so-called ECC process (Error Correction Code), which recognises and corrects memory errors.
Progress and advantages of the development of server RAM
Without the rapid development of the internet and powerful mainframes, our technology today - including AI - would probably not be where it is now. The development of server RAM has brought numerous advantages:
- Greater reliability
ECC RAM enables smooth operation without data corruption and therefore ensures stable systems. - Better scalability
The development of RAM does not stand still - DDR4 and even DDR5 modules now ensure higher speeds with lower energy consumption. - More efficient performance
The development of RAM does not stand still - DDR4 and even DDR5 modules now ensure higher speeds with lower energy consumption. - Optimised for new technologies
More RAM means more ‘power’ - virtual machines can be embedded, which increases the efficiency of data centres and AI systems, for example, can be supplied with the necessary RAM. In addition, modern server RAM enables the processing of complex algorithms in real time, which is particularly essential for applications such as machine learning, big data analyses and cloud computing.
The DDR (Double Data Rate) RAM generations have also evolved significantly over the last few decades and each new version brings more performance, greater efficiency and new technical possibilities. Here is a rough overview:
| Generation | Introducion | Transfer rate (GB/s) | Transfer rate (MT/s) | Voltage | Special feature for servers |
|---|---|---|---|---|---|
| DDR1 | ca. 2000 | 2,1 – 3,2 | 266–400 | 2,5–2,6 V | First ECC-capable server modules |
| DDR2 | ca. 2003 | 4,2 – 6,4 | 533–800 | 1,8 V | Improved bandwidth, stable operation thanks to ECC |
| DDR3 | ca. 2009 | 8,5 – 14,9 | 1066 – 1600 | 1,35 – 1,5 V | Introduction of LRDIMM for higher capacities |
| DDR4 | ca. 2014 | 17 – 25,6 | 2133 – 5100 | 1,2 V | Higher density, better error correction |
| DDR5 | ab 2020 | 38,4 – 51,2 | 3200 – 6400 | 1,1 V | Integrated voltage regulators, double bandwidth, ideal for AI, HPC and cloud |
The constant further development of server RAM is a decisive factor for innovations in almost all digital areas.
We offer a large selection of inexpensive used RAM modules for servers in our hardware shop so that you too can upgrade your servers to more power and performance at low cost.
We have more interesting articles on server RAM and other important server components for you here:
















