Researchers at North Carolina State University claim they've achieved a breakthrough in how encryption can be used in technology called non-volatile main memory, which is seen as an eventual replacement for conventional dynamic random-access memory. Unlike DRAM, with NVMM "when you turn off the power, the data remains there," Yan Solihin, associate professor of electrical and computing engineering at NC State, explains. He notes: "The advantage of that is clear - that's an advantage because it gives you an instant-on experience." But the downside in terms of security is that the user's application data, even as much as 4GB worth, "doesn't disappear from main memory. So if your computer is stolen, someone could easily read out the content of main memory." In work conducted with graduate students, Solihin says NC State researchers completed building a hardware-based method to self- encrypt NVMM data. The idea is it might eventually become integrated into chipsets. "We add an encryption engine to the memory module to encrypt to memory," Solihin says. While software-based encryption has been offered before for NVMM, he believes this is likely the first hardware-based approach to the NVMM question. Encryption is known to greatly slow performance but under some conditions, where just 80 percent of memory storage is encrypted, there's only a 3 percent performance hit with the NC State approach. The remaining 20 percent is encrypted at power-down, he says. In contrast, it's not unusual to see a 50 percent slowdown due to encryption, he adds. The N.C. State project relies on the Advanced Encryption Standard (AES). Solihin says N.C. State is now looking toward how to commercialize its hardware-based encryption approach to NVMM by working with industry, where the hardware-based encryption could also find use in smartphones and other mobile devices.
- Free Whitepaper! Learn how IT is evolving from producer to enabler, and fostering collaboration around analytics.
- Free Whitepaper! Learn how to create an analytics environment that is governed, scalable and self-serve.
- Free Whitepaper! The 5 criteria to help you select the right analytics platform for your organization.