Posted in

How Big Data Has Encouraged a New Memory Technology

Intel Data Center GPU codenamed Crescent Island architectural slide showcasing Xe3P AI optimized GPU IP, up to 480GB LPDDR5x memory capacity, and a 350W air-cooled PCIe form factor.
Intel’s Crescent Island GPU targets AI inference economics by prioritizing massive LPDDR5x memory capacity over costly HBM architectures within a practical 350W air-cooled design.

The emergence of big data has generated a need for more advanced tech, at least when it comes to memory and data storage. Micron a renowned chipmaker has answered the call by announcing their new persistent memory technology. Its an 8GB DDR4 NVDIMM that improves a great deal of performance stats in the way of big data storage, including bandwidth, latency, memory capacity and overall operating cost.

Many companies are now offering cloud storage and big data platforms designed to aggregate and protect sensitive and valuable data. However, no one has been able to tackle some of the biggest challenges that arise when storing said data. To understand, we first need to take a look at how computers handle data storage.

Dynamic Random Access Memory vs. Physical Storage

Dynamic Random Access Memory, or DRAM for short, is what computers use to temporarily store information. It is generally located near the processor and is often used for running processes. The problem with storing information or data in the DRAM is that as soon as the power is shut off, the memory is wiped clean. For scheduled power sessions, this isnt an issue. However, it can be a big problem in the event of a power failure and can sometimes cause data loss.

Data or information that needs to be stored long-term is moved to storage drives, also known as hard drives. They come in many forms, but the most modern and widely used in big data centers are solid state drives, which use flash memory chips to store data. This is because, compared to traditional hard drives SATA or IDE they can be accessed much faster, meaning shorter loading times in most instances.

However, when you compare the two data storage methods, even with a solid state drive, accessing the data stored within is always going to be slower than if it had been in memory. Sadly, its just too dangerous at times to store a large amount of information in the DRAM, especially if it can be lost.

Thats where Microns new NVDIMM chip comes into play. Like DRAM, the new chip sits next to the computer processor and can handle up to 8GB of data at a time. Even better, when or if the power is turned off, the data is not lost.

Plus, it fits just fine into traditional DRAM module slots in existing servers, so theres no need to overhaul existing equipment. This way, if youre already planning something big like transitioning to an IP network theres no need to order specialized equipment other than whats already coming in.

How is this possible, you ask?

How Microns NVDIMM Chip Works

Essentially, Micron merged their DRAM and NAND flash chips so that they would work in tandem. In addition, they were connected to a logic chip that reads incoming data and then tells it where to go, whether that be in long-term storage or memory.

What does this mean?

Tom Eby, Vice President of Microns computer and networking business unit, explains how this will affect current data providers.

Micron is delivering on the promise of persistent memory with a solution that gives system architects a new approach for designing systems with better performance, reduced energy usage and improved total cost of ownership, said Eby. With NVDIMM, we have a powerful solution that is available today. Were also leading the way on future persistent memory development by spearheading R&D efforts on promising new technologies such as 3D XPoint memory, which will be available in 2016 and beyond.

In short, this new technology will enable big data providers to deliver better systems to their customers. Commercial customers can expect to see improved performance, better access and load times and enhanced security for their data.

Micron is even planning to launch a 16GB version of the new NVDIMM chip sometime in the near future. 

Image by SplitShire

Kayla Matthews is a technology writer covering big data, IoT tech and connected technology issues. You can find her other work on ProductivityBytes.com, as well as on Information Age, KDnuggets, The Week and Digital Trends.

Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.