Proceedings of the ACM SIGCOMM Workshop on Network-I/O Convergence: Experience, Lessons, Implications 2003
DOI: 10.1145/944747.944754
|View full text |Cite
|
Sign up to set email alerts
|

A study of iSCSI extensions for RDMA (iSER)

Abstract: The iSCSI protocol is the IETF standard that maps the SCSI family of application protocols onto TCP/IP enabling convergence of storage traffic on to standard TCP/IP fabrics. The ability to efficiently transfer and place the data on TCP/IP networks is crucial for this convergence of the storage traffic. The iWARP protocol suite provides Remote Direct Memory Access (RDMA) semantics over TCP/IP networks and enables efficient memory-to-memory data transfers over an IP fabric. This paper studies the design process … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1

Citation Types

0
3
0

Year Published

2005
2005
2014
2014

Publication Types

Select...
4
2

Relationship

1
5

Authors

Journals

citations
Cited by 23 publications
(3 citation statements)
references
References 3 publications
0
3
0
Order By: Relevance
“…This competition will become more intense in the near future with the advent of the high performance interconnection and transport technique, such as Remote Direct Memory Access(RDMA) [20]. We will target the new condition to rethink the storage management again.…”
Section: Discussionmentioning
confidence: 99%
“…This competition will become more intense in the near future with the advent of the high performance interconnection and transport technique, such as Remote Direct Memory Access(RDMA) [20]. We will target the new condition to rethink the storage management again.…”
Section: Discussionmentioning
confidence: 99%
“…As such, they are used by kernel-based I/O subsystems (e.g., network file systems [5] and storage device drivers [6]), whereas user-level RNICs can be used by either user-level or kernel-based I/O subsystems [13]. In previous work we argued that the performance drawback of using a kernel-based RNIC instead of a userlevel RNIC amounts to the overhead of crossing the userkernel boundary for issuing and managing I/O operations, which is not a significant cost for I/O-intensive network storage workloads [14].…”
Section: Kernel-based Rdmamentioning
confidence: 99%
“…This is a key requirement of storage protocols such as iSCSI and its extensions for RDMA [6] (iSER), where the binding between a memory buffer and the physical memory backing it is not known until the time of I/O. The FAST-REGISTER MEMORY REGION and BIND MEMORY WINDOW APIs developed for this purpose, split the process of registration in two parts; the allocation of NIC resources, such as TLB entries, protection checks, etc.…”
Section: Related Workmentioning
confidence: 99%