Improved Memory Management System for Computer Systems

Back to all technologies
Download as PDF
Data centers share information dictated by network protocols, and in order to operate efficiently, these protocols must access information quickly on a large scale. However, the two leading protocols, Transmission Control Protocol (TCP) and Remote Direct Memory Access (RDMA) both fail in one of those categories: TCP has a high latency and RDMA is not plausible on a large scale.

Researchers at Purdue University have developed a protocol, Remote Indirect Memory Access (RIMA), which manages to have both low latency and works over a large scale. The basic operation for RIMA allows for messages to be appended to the tail of a memory region that both the sender and the receiver can access. Information under RIMA's protocol would only be shared with the two nodes involved in the transfer, not with every single node in the system, which causes RDMA to be difficult to scale. Having the information appended onto the memory regions allows it to still bypass the kernel and maintain a low latency. This eliminates both of the problems that RDMA and TCP experience. RIMA can easily be implemented into existing data centers through a network interface card (NIC) and a software update, revolutionizing large data transfers in data centers.

-Low latency
-Can be installed into existing data centers

Potential Applications:
-Large data centers
-New network protocols
Sep 11, 2018
United States
Sep 28, 2021

Sep 28, 2021
United States

Mar 13, 2017

Mar 11, 2016
United States
Purdue Office of Technology Commercialization
The Convergence Center
101 Foundry Drive, Suite 2500
West Lafayette, IN 47906

Phone: (765) 588-3475
Fax: (765) 463-3486