remote direct memory access (RDMA)
GPTKB entity
Statements (50)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
computer networking
|
| gptkbp:competitor |
TCP/IP networking
|
| gptkbp:enables |
low-latency communication
zero-copy networking low CPU utilization high bandwidth data transfer bypass of operating system kernel direct data placement high efficiency networking remote memory access without CPU intervention direct memory access from memory of one computer to another |
| https://www.w3.org/2000/01/rdf-schema#label |
remote direct memory access (RDMA)
|
| gptkbp:improves |
throughput
|
| gptkbp:introducedIn |
early 2000s
|
| gptkbp:operates |
gptkb:Ethernet
gptkb:InfiniBand gptkb:converged_Ethernet |
| gptkbp:reduces |
CPU involvement in data transfer
latency in data transfer |
| gptkbp:requires |
RDMA-capable network adapters
|
| gptkbp:standardizedBy |
gptkb:IETF
gptkb:InfiniBand_Trade_Association |
| gptkbp:supportedBy |
gptkb:Windows
gptkb:Linux major cloud providers |
| gptkbp:supportsProtocol |
gptkb:RoCE
gptkb:RDMA_over_Converged_Ethernet_(RoCE) gptkb:InfiniBand gptkb:iWARP gptkb:InfiniBand_RDMA gptkb:Internet_Wide_Area_RDMA_Protocol_(iWARP) |
| gptkbp:usedBy |
gptkb:Google_Cloud_Platform
gptkb:Amazon_Web_Services gptkb:Microsoft_Azure |
| gptkbp:usedFor |
big data analytics
distributed databases machine learning workloads storage networking |
| gptkbp:usedIn |
gptkb:NVMe_over_Fabrics
cloud computing data centers high-performance computing industrial equipment scientific computing supercomputers parallel computing financial trading systems distributed file systems |
| gptkbp:bfsParent |
gptkb:NVMe_over_Fabrics
|
| gptkbp:bfsLayer |
6
|