Network File System (NFS) Upper Layer Binding To RPC-Over-RDMA
draft-ietf-nfsv4-rfc5667bis-05
The information below is for an old version of the document.
Document | Type |
This is an older version of an Internet-Draft that was ultimately published as RFC 8267.
|
|
---|---|---|---|
Author | Chuck Lever | ||
Last updated | 2017-02-03 | ||
RFC stream | Internet Engineering Task Force (IETF) | ||
Formats | |||
Reviews | |||
Additional resources | Mailing list discussion | ||
Stream | WG state | WG Document | |
Document shepherd | (None) | ||
IESG | IESG state | Became RFC 8267 (Proposed Standard) | |
Consensus boilerplate | Unknown | ||
Telechat date | (None) | ||
Responsible AD | (None) | ||
Send notices to | (None) |
draft-ietf-nfsv4-rfc5667bis-05
Network Working Group L-E. Jonsson Request for Comments: 3242 G. Pelletier Category: Standards Track Ericsson April 2002 RObust Header Compression (ROHC): A Link-Layer Assisted Profile for IP/UDP/RTP Status of this Memo This document specifies an Internet standards track protocol for the Internet community, and requests discussion and suggestions for improvements. Please refer to the current edition of the "Internet Official Protocol Standards" (STD 1) for the standardization state and status of this protocol. Distribution of this memo is unlimited. Copyright Notice Copyright (C) The Internet Society (2002). All Rights Reserved. Abstract This document defines a ROHC (Robust Header Compression) profile for compression of IP/UDP/RTP (Internet Protocol/User Datagram Protocol/Real-Time Transport Protocol) packets, utilizing functionality provided by the lower layers to increase compression efficiency by completely eliminating the header for most packets during optimal operation. The profile is built as an extension to the ROHC RTP profile. It defines additional mechanisms needed in ROHC, states requirements on the assisting layer to guarantee transparency, and specifies general logic for compression and decompression making use of this header-free packet. Table of Contents 1. Introduction....................................................2 2. Terminology.....................................................4 3. Overview of the Link-Layer Assisted Profile.....................5 3.1. Providing Packet Type Identification.....................6 3.2. Replacing the Sequence Number............................6 3.3. CRC Replacement..........................................7 3.4. Applicability of This Profile............................7 4. Additions and Exceptions Compared to ROHC RTP...................8 4.1. Additional Packet Types..................................8 4.1.1. No-Header Packet (NHP)..........................8 4.1.2. Context Synchronization Packet (CSP)............8 4.1.3. Context Check Packet (CCP)......................9 Jonsson, et. al Standards Track [Page 1] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 4.2. Interfaces Towards the Assisting Layer..................11 4.2.1. Interface, Compressor to Assisting Layer.......11 4.2.2. Interface, Assisting Layer to Decompressor.....12 4.3. Optimistic Approach Agreement...........................13 4.4. Fast Context Initialization, IR Redefinition............13 4.5. Feedback Option, CV-REQUEST.............................14 4.6. Periodic Context Verification...........................15 4.7. Use of Context Identifier...............................15 5. Implementation Issues..........................................15 5.1. Implementation Parameters and Signals...................15 5.1.1. Implementation Parameters at the Compressor....16 5.1.2. Implementation Parameters at the Decompressor..17 5.2. Implementation over Various Link Technologies...........18 6. IANA Considerations............................................18 7. Security Considerations........................................18 8. Acknowledgements...............................................18 9. References.....................................................19 10. Authors' Addresses.............................................20 11. Full Copyright Statement.......................................21 1. Introduction Header compression is a technique used to compress and transparently decompress the header information of a packet on a per-hop basis, utilizing redundancy within individual packets and between consecutive packets within a packet stream. Over the years, several protocols [VJHC, IPHC] have been developed to compress the network and transport protocol headers [IPv4, IPv6, UDP, TCP], and these schemes have been successful in improving efficiency over many wired bottleneck links, such as modem connections over telephone networks. In addition to IP, UDP, and TCP compression, an additional compression scheme called Compressed RTP [CRTP] has been developed to further improve compression efficiency for the case of real-time traffic using the Real-Time Transport Protocol [RTP]. The schemes mentioned above have all been designed taking into account normal assumptions about link characteristics, which traditionally have been based on wired links only. However, with an increasing number of wireless links in the Internet paths, these assumptions are no longer generally valid. In wireless environments, especially wide coverage cellular environments, relatively high error rates are tolerated in order to allow efficient usage of the radio resources. For real-time traffic, which is more sensitive to delays than to errors, such operating conditions will be norm over, for example, 3rd generation cellular links, and header compression must therefore tolerate packet loss. However, with the previously mentioned schemes, especially for real-time traffic compressed by CRTP, high error rates have been shown to significantly degrade Jonsson, et. al Standards Track [Page 2] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 header compression performance [CRTPC]. This problem was the driving force behind the creation of the RObust Header Compression (ROHC) WG in the IETF. The ROHC WG has developed a header compression framework on top of which profiles can be defined for different protocol sets, or for different compression strategies. Due to the limited packet loss robustness of CRTP, and the demands of the cellular industry for an efficient way of transporting voice over IP over wireless, the main focus of ROHC has so far been on compression of IP/UDP/RTP headers, which are generous in size, especially compared to the payloads often carried by packets with such headers. ROHC RTP has become a very efficient, robust and capable compression scheme, able to compress the headers down to a total size of one octet only. Also, transparency is guaranteed to an extremely great extent even when residual bit errors are present in compressed headers delivered to the decompressor. The requirements for RTP compression [RTP-REQ], defined by the WG before and during the development process, have thus been fulfilled. As mentioned above, the 3rd generation cellular systems, where IP will be used end-to-end, have been one of the driving forces behind ROHC RTP, and the scheme has been designed to also suit new cellular air interfaces, such as WCDMA, making it possible to run even speech services with spectrum efficiency insignificantly lower than for existing one-service circuit switched solutions [VTC2000]. However, other air interfaces such as those based on GSM and IS-95 will also be used in all-IP networks, with further implications for the header compression issue. These older air interfaces are less flexible, with radio bearers optimized for specific payload sizes. This means that not even a single octet of header can be added without using the next higher fixed packet size supported by the link, something which is obviously very costly. For the already deployed speech vocoders, the spectrum efficiency over these links will thus be low compared to existing circuit switched solutions. To achieve high spectrum efficiency overall with any application, more flexible air interfaces must be deployed, and then the ROHC RTP scheme will perform excellently, as shown for WCDMA [MOMUC01]. However, for deployment reasons, it is however important to also provide a suitable header compression strategy for already existing vocoders and air interfaces, such as for GERAN and for CDMA2000, with minimal effects on spectral efficiency. This document defines a new link-layer assisted ROHC RTP profile extending ROHC RTP (profile 0x0001) [ROHC], compliant with the ROHC 0-byte requirements [0B-REQ]. The purpose of this new profile is to provide a header-free packet format that, for a certain application Jonsson, et. al Standards Track [Page 3] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 2. Conveying NFS Operations On RPC-Over-RDMA . . . . . . . . . . 3 3. Upper Layer Binding For NFS Versions 2 And 3 . . . . . . . . 4 4. Upper Layer Binding For NFS Version 4 . . . . . . . . . . . . 6 5. Extending NFS Upper Layer Bindings . . . . . . . . . . . . . 12 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 13 7. Security Considerations . . . . . . . . . . . . . . . . . . . 13 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 14 Appendix A. Changes Since RFC 5667 . . . . . . . . . . . . . . . 15 Appendix B. Acknowledgments . . . . . . . . . . . . . . . . . . 17 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 17 1. Introduction An RPC-over-RDMA transport, such as the one defined in [I-D.ietf-nfsv4-rfc5666bis], may employ direct data placement to convey data payloads associated with RPC transactions. To enable successful interoperation, RPC client and server implementations must agree as to which XDR data items in what particular RPC procedures are eligible for direct data placement (DDP). This document contains material required of Upper Layer Bindings, as specified in [I-D.ietf-nfsv4-rfc5666bis], for the following NFS protocol versions: o NFS Version 2 [RFC1094] Lever Expires August 7, 2017 [Page 2] Internet-Draft NFS On RPC-Over-RDMA February 2017 o NFS Version 3 [RFC1813] o NFS Version 4.0 [RFC7530] o NFS Version 4.1 [RFC5661] o NFS Version 4.2 [RFC7862] Upper Layer Bindings specified in this document apply to all versions of RPC-over-RDMA. 2. Conveying NFS Operations On RPC-Over-RDMA Definitions of terminology and a general discussion of how RPC-over- RDMA is used to convey RPC transactions can be found in [I-D.ietf-nfsv4-rfc5666bis]. In this section, these general principles are applied in the context of conveying NFS procedures on RPC-over-RDMA. Some issues common to all NFS protocol versions are introduced. 2.1. DDP Eligibility Violations To report a DDP-eligibity violation, an NFS server MUST return one of: o An RPC-over-RDMA message of type RDMA_ERROR, with the rdma_xid field set to the XID of the matching NFS Call, and the rdma_error field set to ERR_CHUNK; or o An RPC message (via an RDMA_MSG message) with the xid field set to the XID of the matching NFS Call, the mtype field set to REPLY, the stat field set to MSG_ACCEPTED, and the accept_stat field set to GARBAGE_ARGS. Subsequent sections of this document describe further considerations particular to specific NFS protocols or procedures. 2.2. Reply Size Estimation During the construction of each RPC Call message, an NFS client is responsible for allocating appropriate resources for receiving the matching Reply message. A Reply buffer overrun can result in corruption of the Reply message or termination of the transport connection. Therefore reliable reply size estimation is necessary to ensure successful interoperation. This is particularly critical, for example, when allocating a Reply chunk. Lever Expires August 7, 2017 [Page 3] Internet-Draft NFS On RPC-Over-RDMA February 2017 In many cases the Upper Layer Protocol's XDR definition provides enough information to enable the client to make a reliable prediction of the maximum size of the expected Reply message. If there are variable-size data items in the result, the maximum size of the RPC Reply message can be reliably estimated in most cases: o The client requests only a specific portion of an object (for example, using the "count" and "offset" fields in an NFS READ). o The client has already cached the size of the whole object it is about to request (say, via a previous NFS GETATTR request). o The client and server have negotiated a maximum size for all calls and responses. Subsequent sections of this document describe considerations particular to specific NFS procedures where it is not possible to determine the maximum Reply message size based solely on the above criteria. 3. Upper Layer Binding For NFS Versions 2 And 3 This Upper Layer Binding specification applies to NFS Version 2 [RFC1094] and NFS Version 3 [RFC1813]. For brevity, in this section a "legacy NFS client" refers to an NFS client using NFS version 2 or NFS version 3 to communicate with an NFS server. Likewise, a "legacy NFS server" is an NFS server communicating with clients using NFS version 2 or NFS version 3. The following XDR data items in NFS versions 2 and 3 are DDP- eligible: o The opaque file data argument in the NFS WRITE procedure o The pathname argument in the NFS SYMLINK procedure o The opaque file data result in the NFS READ procedure o The pathname result in the NFS READLINK procedure All other argument or result data items in NFS versions 2 and 3 are not DDP-eligible. A legacy server's response to a DDP-eligibility violation (described in Section 2.1) does not give an indication to legacy clients of whether the server has processed the arguments of the RPC Call, or whether the server has accessed or modified client memory associated with that RPC. Lever Expires August 7, 2017 [Page 4] Internet-Draft NFS On RPC-Over-RDMA February 2017 A legacy NFS client determines the maximum reply size for each operation using the basic criteria outlined in Section 2.2. 3.1. Auxiliary Protocols NFS versions 2 and 3 are typically deployed with several other protocols, sometimes referred to as "NFS auxiliary protocols." These are separate RPC programs that define procedures which are not part of the NFS version 2 or version 3 RPC programs. These include: o The MOUNT and NLM protocols, introduced in an appendix of [RFC1813] o The NSM protocol, described in Chapter 11 of [NSM] o The NFSACL protocol, which does not have a public definition (NFSACL here is treated as a de facto standard as there are several interoperating implementations). RPC-over-RDMA considers these programs as distinct Upper Layer Protocols [I-D.ietf-nfsv4-rfc5666bis]. To enable the use of these ULPs on an RPC-over-RDMA transport, an Upper Layer Binding specification is provided here for each. 3.1.1. MOUNT, NLM, And NSM Protocols Typically MOUNT, NLM, and NSM are conveyed via TCP, even in deployments where NFS operations on RPC-over-RDMA. When a legacy server supports these programs on RPC-over-RDMA, it advertises the port address via the usual rpcbind service [RFC1833]. No operation in these protocols conveys a significant data payload, and the size of RPC messages in these protocols is uniformly small. Therefore, no XDR data items in these protocols are DDP-eligible. The largest variable-length XDR data item is an xdr_netobj. In most implementations this data item is not larger than 1024 bytes, making reliable reply size estimation straightforward using the criteria outlined in Section 2.2. 3.1.2. NFSACL Protocol Legacy clients and servers that support the NFSACL RPC program typically convey NFSACL procedures on the same connection as the NFS RPC program. This obviates the need for separate rpcbind queries to discover server support for this RPC program. Lever Expires August 7, 2017 [Page 5] Internet-Draft NFS On RPC-Over-RDMA February 2017 ACLs are typically small, but even large ACLs must be encoded and decoded to some degree. Thus no data item in this Upper Layer Protocol is DDP-eligible. For procedures whose replies do not include an ACL object, the size of a reply is determined directly from the NFSACL program's XDR definition. There is no protocol-wide size limit for NFS version 3 ACLs, and there is no mechanism in either the NFSACL or NFS programs for a legacy client to ascertain the largest ACL a legacy server can store. Legacy client implementations should choose a maximum size for ACLs based on their own internal limits. A recommended lower bound for this maximum is 32,768 bytes. When an especially large ACL is expected, a Reply chunk might be required. If a legacy NFS server indicates that it cannot return an NFSACL GETACL response because the legacy NFS client has not provided a large enough Reply chunk to receive that response, the legacy NFS client can choose to o Terminate the NFSACL GETACL with an error, or o Allocate a larger Reply chunk and send the same NFSACL GETACL request as a new RPC transaction. The NFS client should avoid retrying the request indefinitely. 4. Upper Layer Binding For NFS Version 4 This Upper Layer Binding specification applies to all protocols defined in NFS Version 4.0 [RFC7530], NFS Version 4.1 [RFC5661], and NFS Version 4.2 [RFC7862]. 4.1. DDP-Eligibility Only the following XDR data items in the COMPOUND procedure of all NFS version 4 minor versions are DDP-eligible: o The opaque data field in the WRITE4args structure o The linkdata field of the NF4LNK arm in the createtype4 union o The opaque data field in the READ4resok structure o The linkdata field in the READLINK4resok structure Lever Expires August 7, 2017 [Page 6] Internet-Draft NFS On RPC-Over-RDMA February 2017 o In minor version 2 and newer, the rpc_data field of the read_plus_content union (further restrictions on the use of this data item follow below). 4.1.1. READ_PLUS Replies The NFS version 4.2 READ_PLUS operation returns a complex data type [RFC7862]. The rpr_contents field in the result of this operation is an array of read_plus_content unions, one arm of which contains an opaque byte stream (d_data). The size of d_data is limited to the value of the rpa_count field, but the protocol does not bound the number of elements which can be returned in the rpr_contents array. In order to make the size of READ_PLUS replies predictable by NFS version 4.2 clients, the following restrictions are placed on the use of the READ_PLUS operation on RPC-over-RDMA transports: o An NFS version 4.2 client MUST NOT provide more than one Write chunk for any READ_PLUS operation. When providing a Write chunk for a READ_PLUS operation, an NFS version 4.2 client MUST provide a Write chunk that is either empty (which forces all result data items for this operation to be returned inline) or large enough to receive rpa_count bytes in a single element of the rpr_contents array. o If the Write chunk provided for a READ_PLUS operation by an NFS version 4.2 client is not empty, an NFS version 4.2 server MUST use that chunk for the first element of the rpr_contents array that has an rpc_data arm. o An NFS version 4.2 server MUST NOT return more than two elements in the rpr_contents array of any READ_PLUS operation. It returns as much of the requested byte range as it can fit within these two elements. If the NFS version 4.2 server has not asserted rpr_eof in the reply, the NFS version 4.2 client SHOULD send additional READ_PLUS requests for any remaining bytes. 4.2. NFS Version 4 Reply Size Estimation Within NFS version 4, there are certain variable-length result data items whose maximum size cannot be estimated by clients reliably because there is no protocol-specified size limit on these arrays. These include: o The attrlist4 field Lever Expires August 7, 2017 [Page 7] Internet-Draft NFS On RPC-Over-RDMA February 2017 o Fields containing ACLs such as fattr4_acl, fattr4_dacl, fattr4_sacl o Fields in the fs_locations4 and fs_locations_info4 data structures o Fields opaque to the NFS version 4 protocol which pertain to pNFS layout metadata, such as loc_body, loh_body, da_addr_body, lou_body, lrf_body, fattr_layout_types and fs_layout_types, 4.2.1. Reply Size Estimation For Minor Version 0 The NFSv4.0 protocol itself does not impose any bound on the size of NFS calls or responses. Some of the data items enumerated in Section 4.2 (in particular, the items related to ACLs and fs_locations) make it difficult to predict the maximum size of NFSv4.0 GETATTR replies that interrogate variable-length attributes. As discussed in Section 2.2, client implementations can rely on their own internal architectural limits to bound the reply size, but such limits are not always guaranteed to be reliable. When an especially large NFSv4.0 GETATTR result is expected, a Reply chunk might be required. If an NFSv4.0 server indicates that it cannot return an NFSv4.0 GETATTR response because the requesting NFSv4.0 client has not provided a large enough Reply chunk to receive that response, the NFSv4.0 client can choose to o Terminate the NFSv4.0 GETATTR with an error, or o Allocate a larger Reply chunk and send the same NFSv4.0 GETATTR request as a new RPC transaction. The NFS client should avoid retrying the request indefinitely. The use of NFS COMPOUND operations raises the possibility of requests that combine a non-idempotent operation (eg. NFS WRITE) with an NFSv4.0 GETATTR that requests one or more variable length results. This combination should be avoided by ensuring that any NFSv4.0 GETATTR operation that might return a result of unpredictable length is sent in an NFS COMPOUND by itself. 4.2.2. Reply Size Estimation For Minor Version 1 And Newer In NFS version 4.1 and newer minor versions, the csa_fore_chan_attrs argument of the CREATE_SESSION operation contains a ca_maxresponsesize field. The value in this field can be taken as the absolute maximum size of replies generated by a replying NFS version 4 server. Lever Expires August 7, 2017 [Page 8] Internet-Draft NFS On RPC-Over-RDMA February 2017 This value can be used in cases where it is not possible to estimate a reply size upper bound precisely. In practice, objects such as ACLs, named attributes, layout bodies, and security labels are much smaller than this maximum. 4.3. NFS Version 4 COMPOUND Requests The NFS version 4 COMPOUND procedure allows the transmission of more than one DDP-eligible data item per Call and Reply message. An NFS version 4 client provides XDR Position values in each Read chunk to disambiguate which chunk is associated with which argument data item. However NFS version 4 server and client implementations must agree in advance on how to pair Write chunks with returned result data items. The mechanism specified in Section 4.3.2 of [I-D.ietf-nfsv4-rfc5666bis]) is applied here, with additional restrictions that appear below. In the following list, an "NFS Read" operation refers to any NFS Version 4 operation which has a DDP- eligible result data item (i.e., either a READ, READ_PLUS, or READLINK operation). o If an NFS version 4 client wishes all DDP-eligible items in an NFS reply to be conveyed inline, it leaves the Write list empty. o The first chunk in the Write list MUST be used by the first READ operation in an NFS version 4 COMPOUND procedure. The next Write chunk is used by the next READ operation, and so on. o If an NFS version 4 client has provided a matching non-empty Write chunk, then the corresponding READ operation MUST return its DDP- eligible data item using that chunk. o If an NFS version 4 client has provided an empty matching Write chunk, then the corresponding READ operation MUST return all of its result data items inline. o If a READ operation returns a union arm which does not contain a DDP-eligible result, and the NFS version 4 client has provided a matching non-empty Write chunk, an NFS version 4 server MUST return an empty Write chunk in that Write list position. o If there are more READ operations than Write chunks, then remaining NFS Read operations in an NFS version 4 COMPOUND that have no matching Write chunk MUST return their results inline. Lever Expires August 7, 2017 [Page 9] Internet-Draft NFS On RPC-Over-RDMA February 2017 4.3.1. NFS Version 4 COMPOUND Example The following example shows a Write list with three Write chunks, A, B, and C. The NFS version 4 server consumes the provided Write chunks by writing the results of the designated operations in the compound request (READ and READLINK) back to each chunk. Write list: A --> B --> C NFS version 4 COMPOUND request: PUTFH LOOKUP READ PUTFH LOOKUP READLINK PUTFH LOOKUP READ | | | v v v A B C If the NFS version 4 client does not want to have the READLINK result returned via RDMA, it provides an empty Write chunk for buffer B to indicate that the READLINK result must be returned inline. 4.4. NFS Version 4 Callback The NFS version 4 protocols support server-initiated callbacks to notify clients of events such as recalled delegations. 4.4.1. NFS Version 4.0 Callback NFS version 4.0 implementations typically employ a separate TCP connection to handle callback operations, even when the forward channel uses a RPC-over-RDMA transport. No operation in the NFS version 4.0 callback RPC program conveys a significant data payload. Therefore, no XDR data items in this RPC program is DDP-eligible. A CB_RECALL reply is small and fixed in size. The CB_GETATTR reply contains a variable-length fattr4 data item. See Section 4.2.1 for a discussion of reply size prediction for this data item. An NFS version 4.0 client advertises netids and ad hoc port addresses for contacting its NFS version 4.0 callback service using the SETCLIENTID operation. Lever Expires August 7, 2017 [Page 10] behavior, can replace a majority of the 1-octet header ROHC RTP packets during normal U/O-mode operation, while still being fully transparent and complying with all the requirements of ROHC RTP [RTP-REQ]. For other applications, compression will be carried out as with normal ROHC RTP. To completely eliminate the compressed header, all functionality normally provided by the 1-octet header has to be provided by other means, typically by utilizing functionality provided by the lower layers and sacrificing efficiency for less frequently occurring larger compressed headers. The latter is not a contradiction since the argument for eliminating the last octet for most packets is not overall efficiency in general. It is important to remember that the purpose of this profile is to provide efficient matching of existing applications to existing link technologies, not efficiency in general. The additional complexity introduced by this profile, although minimized by a tight integration with already existing ROHC functionality, implies that it should therefore only be used to optimize performance of specific applications over specific links. When implementing this profile over various link technologies, care must be taken to guarantee that all the functionality needed is provided by ROHC and the lower layers together. Therefore, additional documents should specify how to incorporate this profile on top of various link technologies. 2. Terminology The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. CCP Context Check Packet CRC Cyclic Redundancy Check CSP Context Synchronization Packet LLA Link Layer Assisted ROHC RTP profile NHP No Header Packet ROHC RObust Header Compression RHP ROHC Header Packet (a non-NHP packet, i.e., RRP, CSP or CCP) RRP ROHC RTP Packet as defined in [ROHC, profile 0x0001] Assisting layer "Assisting layer" refers to any entity implementing the interface to ROHC (section 4.2). It may, for example, refer to a sub-layer used to adapt the ROHC implementation and the physical link layer. This layer is assumed to have knowledge of the physical layer synchronization. Jonsson, et. al Standards Track [Page 4] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 Compressing side "Compressing side" refers to the combination of the header compressor, operating with the LLA profile, and its associated assisting layer. Lower layers "Lower layers" in this document refers to entities located below ROHC in the protocol stack, including the assisting layer. ROHC RTP "ROHC RTP" in this document refers to the IP/UDP/RTP profile (profile 0x0001) as defined in [ROHC]. 3. Overview of the Link-Layer Assisted Profile The ROHC IP/UDP/RTP profile defined in this document, profile 0x0005 (hex), is designed to be used over channels that have been optimized for specific payload sizes and therefore cannot efficiently accommodate header information when transmitted together with payloads corresponding to these optimal sizes. The LLA profile extends, and thus also inherits all functionality from, the ROCH RTP profile by defining some additional functionality and an interface from the ROHC component towards an assisting lower layer. +---------------------------------------+ | | The LLA | ROHC RTP, | profile | Profile #1 +-----------------+ | | LLA Additions | +---------------------+-----------------+ By imposing additional requirements on the lower layers compared to [ROHC], it is possible to infer the information needed to maintain robust and transparent header compression even though the headers are completely eliminated during most of the operation time. Basically, what this profile does is to replace the smallest and most frequent ROHC U/O-mode headers with a no-header format, for which the header functionality must be provided by other means. Jonsson, et. al Standards Track [Page 5] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 Smallest header in Smallest header in ROHC RTP (profile #1) LLA (profile #5) +--+--+--+--+--+--+--+--+ ++ | 1 octet | -----> || No Header +--+--+--+--+--+--+--+--+ ++ | | Header field functionality +-------------------> provided by other means The fields present in the ROHC RTP headers for U/O-mode PT0 are the packet type identifier, the sequence number and the CRC. The subsequent sections elaborate more on how the functionality of these fields is replaced for NHP. 3.1. Providing Packet Type Identification All ROHC headers carry a packet type identifier, indicating to the decompressor how the header should be interpreted. This is a function that must be provided by some means in 0-byte header compression. ROHC RTP packets with compressed headers will be possible to distinguish thanks to the packet type identifier, but a mechanism is needed to separate packets with a header from packets without a header. This function MUST therefore be provided by the assisting layer in one way or another. 3.2. Replacing the Sequence Number From the sending application, the RTP sequence number is increased by one for each packet sent. The purpose of the sequence number is to cope with packet reordering and packet loss. If reordering or loss has occurred before the transmission point, if needed the compressing side can easily avoid problems by not allowing the use of a header- free packet. However, at the transmission point, loss or reordering that may occur over the link can not be anticipated and covered for. Therefore, for NHP the assisting layer MUST guarantee in-order delivery over the link (already assumed by [ROHC]) and at the receiving side it MUST provide an indication for each packet loss over the link. This is basically the same principle as the VJ header compression [VJHC] relies on. Note that guaranteeing in-order delivery and packet loss indication over the link not only makes it possible to infer the sequence number information, but also supersedes the main function of the CRC, which normally takes care of errors due to long link losses and bit errors in the compressed sequence number. Jonsson, et. al Standards Track [Page 6] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 3.3. CRC Replacement All context updating RRP packets carry a CRC calculated over the uncompressed header. The CRC is used by the decompressor to verify that the updated context is correct. This verification serves three purposes in U/O-mode: 1) Detection of longer losses than can be covered by the sequence number LSBs 2) Protection against failures caused by residual bit errors in compressed headers 3) Protection against faulty implementations and other causes of error Since this profile defines an NHP packet without this CRC, care must be taken to fulfill these purposes by other means, when an NHP is used as a replacement for a context updating packet. Detection of long losses (1) is already covered since the assisting layer MUST provide indication of all packet losses. Furthermore, the NHP packet has one important advantage over RHP packets in that residual bit errors (2) cannot damage a header that is not even sent. It is thus reasonable to assume that compression and decompression transparency can be assured with high confidence even without a CRC in header-free packets. However, to provide additional protection against damage propagation due to undetected residual bit errors in context updating packets (2) or other unexpected errors (3), periodic context verifications SHOULD be performed (see section 4.6). 3.4. Applicability of This Profile The LLA profile can be used with any link technology capable of providing the required functionality described in previous sections. Whether LLA or ROHC RTP should be implemented thus depends on the characteristics of the link itself. For most RTP packet streams, LLA will work exactly as ROHC RTP, while it will be more efficient for packet streams with certain characteristics. LLA will never be less efficient than ROHC RTP. Note as well that LLA, like all other ROHC profiles, is fully transparent to any packet stream reaching the compressor. LLA does not make any assumptions about the packet stream but will perform optimally for packet streams with certain characteristics, e.g., synchronized streams exactly timed with the assisting link over which the LLA profile is implemented. Jonsson, et. al Standards Track [Page 7] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 The LLA profile is obviously not applicable if the UDP checksum (2 bytes) is enabled, which is always the case for IPv6/UDP. For IPv4/UDP, the sender may choose to disable the UDP checksum. 4. Additions and Exceptions Compared to ROHC RTP 4.1. Additional Packet Types The LLA profile defines three new packet types to be used in addition to the RRP packet types defined by [ROHC]. The following sections describe these packet types and their purpose in detail. 4.1.1. No-Header Packet (NHP) A No-Header Packet (NHP), i.e., a packet consisting only of a payload, is defined and MAY be used when only sequencing must be conveyed, i.e., when all header fields are either unchanged or follow the currently established change pattern. In addition, there are some considerations for the use of the NHP (see 4.3, 4.5 and 4.6). An LLA compressor is not allowed to deliver NHP packets when operating in R-mode. The assisting layer MAY send the NHP for RTP SN = X only if an NHP was delivered by the LLA compressor AND the assisting layer can guarantee that the decompressor will infer the proper sequencing for this NHP. This guarantee is based on the confidence that the decompressor a) has the means to infer proper sequencing for the packet corresponding to SN = X-1, AND b) has either received a loss indication or the packet itself for the packet corresponding to SN = X-1. Updating properties: NHP packets update context (RTP Sequence Number). 4.1.2. Context Synchronization Packet (CSP) The case where the packet stream overruns the channel bandwidth may lead to data being discarded, which may result in decompressor context invalidation. It might therefore be beneficial to send a packet with only the header information and discard the payload. This would be helpful to maintain synchronization of the decompressor context, while efficiently using the available bandwidth. Jonsson, et. al Standards Track [Page 8] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 This case can be handled with the Context Synchronization Packet (CSP), which has the following format: 0 1 2 3 4 5 6 7 +---+---+---+---+---+---+---+---+ | 1 1 1 1 1 0 1 0 | Packet type identifier +===+===+===+===+===+===+===+===+ : ROHC header without padding : : see [ROHC, section 5.7] : +---+---+---+---+---+---+---+---+ Updating properties: CSP maintains the updating properties of the ROHC header it carries. The CSP is defined by one of the unused packet type identifiers from ROHC RTP, carried in the one-octet base header. As for any ROHC packet, except the NHP, the packet may begin with ROHC padding and/or feedback. It may also carry context identification after the packet type identifier. It is possible to have two CID fields present, one after the packet type ID and one within the encapsulated ROHC header. If a decompressor receives a CSP with two non-equal CID values included, the packet MUST be discarded. ROHC segmentation may also be applied to the CSP. Note that when the decompressor has received and processed a CSP, the packet (including any possible data following the CSP encapsulated compressed header) MUST be discarded. 4.1.3. Context Check Packet (CCP) A Context Check Packet (CCP), which does not carry any payload but only an optional CRC value in addition to the packet type identifier, is defined. The purpose of the CCP is to provide a useful packet that MAY be sent by a synchronized physical link layer in the case where data must be sent at fixed intervals, even if no compressed packet is available. Whether the CCP is sent over the link and delivered to the decompressor is decided by the assisting layer. The CCP has the following format: Jonsson, et. al Standards Track [Page 9] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 0 1 2 3 4 5 6 7 +---+---+---+---+---+---+---+---+ | 1 1 1 1 1 0 1 1 | Packet type identifier +===+===+===+===+===+===+===+===+ | C | CRC | +---+---+---+---+---+---+---+---+ C: C = 0 indicates that the CRC field is not used; C = 1 indicates that a valid CRC is present. Updating properties: CCP packets do not update context. The CCP is defined by one of the unused packet type identifiers from ROHC RTP, carried in the first octet of the base header. The first bit of the second octet, the C bit, indicates whether the CRC field is used or not. If C=1, the CRC field MUST be set to the 7-bits CRC calculated over the original uncompressed header defined in [ROHC section 5.9.2]. As for any ROHC packet, except NHP, the packet MAY begin with ROHC padding and/or carry context identification. The use of the CRC field to perform decompressor context verification is optional and is therefore a compressor implementation issue. However, a CCP MUST always be made available to the assisting layer. If the assisting layer receives CCPs with the C-bit set (C=1) from the compressor, it MUST use the last CCP received if a CCP is to be sent, i.e., the CCP corresponding to the last non-CCP packet sent (NHP, RRP or CSP). An assisting layer MAY use the CCP for other purposes, such as signaling a packet loss before the link. The decompressor is REQUIRED to handle a CCP received with the C bit set (C=1), indicating a valid CRC field, and perform context verification. The received CRC MUST then be applied to the last decompressed packet, unless a packet loss indication was previously received. Upon CRC failure, actions MUST be taken as specified in [ROHC, section 5.3.2.2.3]. A CCP received with C=0 MUST be ignored by the decompressor. The decompressor is not allowed to make any further interpretation of the CCP. The use of CCP by an assisting layer is optional and depends on the characteristics of the actual link. Whether it is used or not MUST therefore be specified in link layer implementation specifications for this profile. Jonsson, et. al Standards Track [Page 10] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 4.2. Interfaces Towards the Assisting Layer This profile relies on the lower layers to provide the necessary functionality to allow NHP packets to be sent. This interaction between LLA and the assisting layer is defined as interfaces between the LLA compressor/decompressor and the LLA applicable link technology. | | + + +-------------------------+ +-------------------------+ | ROHC RTP HC | | ROHC RTP HD | +-------------------------+ +-------------------------+ | LLA profile | | LLA profile | +=========================+ +=========================+ | Interface | | Interface | | ROHC to assisting layer | | Assisting layer to ROHC | +=========================+ +=========================+ | Applicable | | Applicable | | link technology | | link technology | +=========================+ +=========================+ | | +------>---- CHANNEL ---->-----+ The figure above shows the various levels, as defined in [ROHC] and this document, constituting a complete implementation of the LLA profile. The figure also underlines the need for additional documents to specify how to implement these interfaces for a link technology for which this profile is relevant. This section defines the information to be exchanged between the LLA compressor and the assisting layer for this profile to operate properly. While it does define semantics, it does not specify how these interfaces are to be implemented. 4.2.1. Interface, Compressor to Assisting Layer This section defines the interface semantics between the compressor and the assisting layer, providing rules for packet delivery from the compressor. The interface defines the following parameters: RRP, RRP segmentation flag, CSP, CSP segmentation flag, NHP, and RTP Sequence Number. All parameters, except the NHP, MUST always be delivered to the assisting layer. This leads to two possible delivery scenarios: a. RRP, CSP, CCP, NHP and RTP Sequence Number are delivered, along with the corresponding segmentation flags set accordingly. Jonsson, et. al Standards Track [Page 11] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002 This corresponds to the case when the compressor allows sending of an NHP packet, with or without segmentation being applied to the corresponding RRP/CSP packets. Recall that delivery of an NHP packet occurs when the ROHC RTP compressor would have used a ROHC UO-0. b. RRP, CSP, CCP and RTP Sequence Number are delivered, along with the corresponding segmentation flags set accordingly. This corresponds to the case when the compressor does not allow sending of an NHP packet. Segmentation might be applied to the corresponding RRP and CSP packets. Segmentation may be applied independently to an RRP or a CSP packet if its size exceeds the largest value provided in the PREFERRED PACKET_SIZES list and if the LARGE_PACKET_ALLOWED parameter is set to false. The segmentation flags are explicitly stated in the interface definition to emphasize that the RRP and the CSP may be delivered by the compressor as segmented packets. The RTP SN MUST be delivered for each packet by the compressor to allow the assisting layer to maintain the necessary sequencing information. 4.2.2. Interface, Assisting Layer to Decompressor Here the interface semantics between the assisting layer and the decompressor are defined, providing simple rules for the delivery of received packets to the decompressor. The decompressor needs a way to distinguish NHP packets from RHP packets. Also, when receiving packets without a header, the decompressor needs a way to infer the sequencing information to keep synchronization between the received payload and the sequence information of the decompressed headers. To achieve this, the decompressor MUST receive the following from the assisting layer: - an indication for each packet loss over the link between the compressing and decompressing sides for CID=0 - the received packet together with an indication whether the packet received is an NHP or not Note that the context is updated from a packet loss indication. Jonsson, et. al Standards Track [Page 12] Internet-Draft NFS On RPC-Over-RDMA February 2017 4.4.2. NFS Version 4.1 Callback In NFS version 4.1 and newer minor versions, callback operations may appear on the same connection as is used for NFS version 4 forward channel client requests. NFS version 4 clients and servers MUST use the mechanism described in [I-D.ietf-nfsv4-rpcrdma-bidirection] when backchannel operations are conveyed on RPC-over-RDMA transports. The csa_back_chan_attrs argument of the CREATE_SESSION operation contains a ca_maxresponsesize field. The value in this field can be taken as the absolute maximum size of backchannel replies generated by a replying NFS version 4 client. There are no DDP-eligible data items in callback procedures defined in NFS version 4.1 or NFS version 4.2. However, some callback operations, such as messages that convey device ID information, can be large, in which case a Long Call or Reply might be required. When an NFS version 4.1 client can support Long Calls in its backchannel, it reports a backchannel ca_maxrequestsize that is larger than the connection's inline thresholds. Otherwise an NFS version 4 server MUST use only Short messages to convey backchannel operations. 4.5. Session-Related Considerations Typically the presence of an NFS session [RFC5661] has no effect on the operation of RPC-over-RDMA. None of the operations introduced to support NFS sessions (eg. SEQUENCE) contain DDP-eligible data items. There is no need to match the number of session slots with the number of available RPC-over-RDMA credits. When an NFS session operates on an RPC-over-RDMA transport, there are a few additional cases where an RPC transaction can fail. For example, a requester might receive, in response to an RPC request, an RDMA_ERROR message with an rdma_err value of ERR_CHUNK, or an RDMA_MSG containing an RPC_GARBAGEARGS reply. These situations are no different from existing RPC errors which an NFS session implementation is already prepared to handle for other transports. As with other transports during such a failure, there might be no SEQUENCE result available to the requester to distinguish whether failure occurred before or after the requested operations were executed on the responder. When a transport error occurs (eg. RDMA_ERROR), the requester proceeds as usual to match the incoming XID value to a waiting RPC Call. The RPC transaction is terminated, and the result status is reported to the Upper Layer Protocol. The requester's session implementation then determines the session ID and Lever Expires August 7, 2017 [Page 11] Internet-Draft NFS On RPC-Over-RDMA February 2017 slot for the failed request, and performs slot recovery to make that slot usable again. If this is not done, that slot could be rendered permanently unavailable. 4.6. Retransmission And Keep-Alive NFS version 4 client implementations often rely on a transport-layer keep-alive mechanism to detect when an NFS version 4 server has become unresponsive. When an NFS server is no longer responsive, client-side keep-alive terminates the connection, which in turn triggers reconnection and RPC retransmission. Some RDMA transports (such as Reliable Connections on InfiniBand) have no keep-alive mechanism. Without a disconnect or new RPC traffic, such connections can remain alive long after an NFS server has become unresponsive. Once an NFS client has consumed all available RPC-over-RDMA credits on that transport connection, it will forever await a reply before sending another RPC request. NFS version 4 clients SHOULD reserve one RPC-over-RDMA credit to use for periodic server or connection health assessment. This credit can be used to drive an RPC request on an otherwise idle connection, triggering either a quick affirmative server response or immediate connection termination. In addition to network partition and request loss scenarios, RPC- over-RDMA connections can be terminated when a Transport header is malformed, messages are larger than receive resources, or when too many RPC-over-RDMA messages are sent at once. In such cases: o If there is a transport error indicated (ie, RDMA_ERROR) before the disconnect or instead of a disconnect, the requester MUST respond to that error as prescribed by the specification of the RPC transport. Then the NFS version 4 rules for handling retransmission apply. o If there is a transport disconnect and the responder has provided no other response for a request, then only the NFS version 4 rules for handling retransmission apply. 5. Extending NFS Upper Layer Bindings RPC programs such as NFS are required to have an Upper Layer Binding specification to interoperate on RPC-over-RDMA transports [I-D.ietf-nfsv4-rfc5666bis]. Via standards action, the Upper Layer Binding specified in this document can be extended to cover versions of the NFS version 4 protocol specified after NFS version 4 minor Lever Expires August 7, 2017 [Page 12] Internet-Draft NFS On RPC-Over-RDMA February 2017 version 2, or separately published extensions to an existing NFS version 4 minor version, as described in [I-D.ietf-nfsv4-versioning]. 6. IANA Considerations NFS use of direct data placement introduces a need for an additional NFS port number assignment for networks that share traditional UDP and TCP port spaces with RDMA services. The iWARP [RFC5041] [RFC5040] protocol is such an example (InfiniBand is not). NFS servers for versions 2 and 3 [RFC1094] [RFC1813] traditionally listen for clients on UDP and TCP port 2049, and additionally, they register these with the portmapper and/or rpcbind [RFC1833] service. However, [RFC7530] requires NFS version 4 servers to listen on TCP port 2049, and they are not required to register. An NFS version 2 or version 3 server supporting RPC-over-RDMA on such a network and registering itself with the RPC portmapper MAY choose an arbitrary port, or MAY use the alternative well-known port number for its RPC-over-RDMA service. The chosen port MAY be registered with the RPC portmapper under the netid assigned by the requirement in [I-D.ietf-nfsv4-rfc5666bis]. An NFS version 4 server supporting RPC-over-RDMA on such a network MUST use the alternative well-known port number for its RPC-over-RDMA service. Clients SHOULD connect to this well-known port without consulting the RPC portmapper (as for NFS version 4 on TCP transports). The port number assigned to an NFS service over an RPC-over-RDMA transport is available from the IANA port registry [RFC3232]. 7. Security Considerations RPC-over-RDMA supports all RPC security models, including RPCSEC_GSS security and transport-level security [RFC2203]. The choice of what Direct Data Placement mechanism to convey RPC argument and results does not affect this, since it changes only the method of data transfer. Specifically, the requirements of [I-D.ietf-nfsv4-rfc5666bis] ensure that this choice does not introduce new vulnerabilities. Because this document defines only the binding of the NFS protocols atop [I-D.ietf-nfsv4-rfc5666bis], all relevant security considerations are therefore to be described at that layer. Lever Expires August 7, 2017 [Page 13] Internet-Draft NFS On RPC-Over-RDMA February 2017 8. References 8.1. Normative References [I-D.ietf-nfsv4-rfc5666bis] Lever, C., Simpson, W., and T. Talpey, "Remote Direct Memory Access Transport for Remote Procedure Call, Version One", draft-ietf-nfsv4-rfc5666bis-09 (work in progress), January 2017. [I-D.ietf-nfsv4-rpcrdma-bidirection] Lever, C., "Bi-directional Remote Procedure Call On RPC- over-RDMA Transports", draft-ietf-nfsv4-rpcrdma- bidirection-06 (work in progress), January 2017. [RFC1833] Srinivasan, R., "Binding Protocols for ONC RPC Version 2", RFC 1833, DOI 10.17487/RFC1833, August 1995, <http://www.rfc-editor.org/info/rfc1833>. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997, <http://www.rfc-editor.org/info/rfc2119>. [RFC2203] Eisler, M., Chiu, A., and L. Ling, "RPCSEC_GSS Protocol Specification", RFC 2203, DOI 10.17487/RFC2203, September 1997, <http://www.rfc-editor.org/info/rfc2203>. [RFC5661] Shepler, S., Ed., Eisler, M., Ed., and D. Noveck, Ed., "Network File System (NFS) Version 4 Minor Version 1 Protocol", RFC 5661, DOI 10.17487/RFC5661, January 2010, <http://www.rfc-editor.org/info/rfc5661>. [RFC7530] Haynes, T., Ed. and D. Noveck, Ed., "Network File System (NFS) Version 4 Protocol", RFC 7530, DOI 10.17487/RFC7530, March 2015, <http://www.rfc-editor.org/info/rfc7530>. [RFC7862] Haynes, T., "Network File System (NFS) Version 4 Minor Version 2 Protocol", RFC 7862, DOI 10.17487/RFC7862, November 2016, <http://www.rfc-editor.org/info/rfc7862>. 8.2. Informative References [I-D.ietf-nfsv4-versioning] Noveck, D., "Rules for NFSv4 Extensions and Minor Versions", draft-ietf-nfsv4-versioning-09 (work in progress), December 2016. Lever Expires August 7, 2017 [Page 14] Internet-Draft NFS On RPC-Over-RDMA February 2017 [NSM] The Open Group, "Protocols for Interworking: XNFS, Version 3W", February 1998. [RFC1094] Nowicki, B., "NFS: Network File System Protocol specification", RFC 1094, DOI 10.17487/RFC1094, March 1989, <http://www.rfc-editor.org/info/rfc1094>. [RFC1813] Callaghan, B., Pawlowski, B., and P. Staubach, "NFS Version 3 Protocol Specification", RFC 1813, DOI 10.17487/RFC1813, June 1995, <http://www.rfc-editor.org/info/rfc1813>. [RFC3232] Reynolds, J., Ed., "Assigned Numbers: RFC 1700 is Replaced by an On-line Database", RFC 3232, DOI 10.17487/RFC3232, January 2002, <http://www.rfc-editor.org/info/rfc3232>. [RFC5040] Recio, R., Metzler, B., Culley, P., Hilland, J., and D. Garcia, "A Remote Direct Memory Access Protocol Specification", RFC 5040, DOI 10.17487/RFC5040, October 2007, <http://www.rfc-editor.org/info/rfc5040>. [RFC5041] Shah, H., Pinkerton, J., Recio, R., and P. Culley, "Direct Data Placement over Reliable Transports", RFC 5041, DOI 10.17487/RFC5041, October 2007, <http://www.rfc-editor.org/info/rfc5041>. [RFC5667] Talpey, T. and B. Callaghan, "Network File System (NFS) Direct Data Placement", RFC 5667, DOI 10.17487/RFC5667, January 2010, <http://www.rfc-editor.org/info/rfc5667>. Appendix A. Changes Since RFC 5667 Corrections and updates made necessary by new language in [I-D.ietf-nfsv4-rfc5666bis] have been introduced. For example, references to deprecated features of RPC-over-RDMA Version One, such as RDMA_MSGP, and the use of the Read list for handling RPC replies, have been removed. The term "mapping" has been replaced with the term "binding" or "Upper Layer Binding" throughout the document. Some material that duplicates what is in [I-D.ietf-nfsv4-rfc5666bis] has been deleted. Material required by [I-D.ietf-nfsv4-rfc5666bis] for Upper Layer Bindings that was not present in [RFC5667] has been added, including discussion of how each NFS version properly estimates the maximum size of RPC replies. Technical corrections have been made. For example, the mention of 12KB and 36KB inline thresholds have been removed. The reference to Lever Expires August 7, 2017 [Page 15] Internet-Draft NFS On RPC-Over-RDMA February 2017 a non-existant NFS version 4 SYMLINK operation has been replaced with NFS version 4 CREATE(NF4LNK). The discussion of NFS version 4 COMPOUND handling has been completed. Some changes were made to the algorithm for matching DDP-eligible results to Write chunks. Requirements to ignore extra Read or Write chunks have been removed from the NFS version 2 and 3 Upper Layer Binding, as they conflict with [I-D.ietf-nfsv4-rfc5666bis]. A complete discussion of reply size estimation has been introduced for all protocols covered by the Upper Layer Bindings in this document. A section discussing NFS version 4 retransmission and connection loss has been added. The following additional improvements have been made, relative to [RFC5667]: o An explicit discussion of NFS version 4.0 and NFS version 4.1 backchannel operation has replaced the previous treatment of callback operations. o A binding for NFS version 4.2 has been added that includes discussion of new data-bearing operations like READ_PLUS. o A section suggesting a mechanism for periodically assessing connection health has been introduced. o Language inconsistent with or contradictory to [I-D.ietf-nfsv4-rfc5666bis] has been removed from Sections 2 and 3, and both Sections have been combined into Section 2 in the present document. o Ambiguous or erroneous uses of RFC2119 terms have been corrected. o References to obsolete RFCs have been updated. o An IANA Considerations Section has replaced the "Port Usage Considerations" Section. o Code excerpts have been removed, and figures have been modernized. Lever Expires August 7, 2017 [Page 16] Internet-Draft NFS On RPC-Over-RDMA February 2017 Appendix B. Acknowledgments The author gratefully acknowledges the work of Brent Callaghan and Tom Talpey on the original NFS Direct Data Placement specification [RFC5667]. The author also wishes to thank Bill Baker and Greg Marsden for their support of this work. Dave Noveck provided excellent review, constructive suggestions, and consistent navigational guidance throughout the process of drafting this document. Dave also contributed the text of Section 4.5 Thanks to Karen Deitke for her sharp observations about idempotency, and the clarity of the discussion of NFS COMPOUNDs and NFS sessions. Special thanks go to Transport Area Director Spencer Dawkins, nfsv4 Working Group Chair Spencer Shepler, and nfsv4 Working Group Secretary Thomas Haynes for their support. Author's Address Charles Lever (editor) Oracle Corporation 1015 Granger Avenue Ann Arbor, MI 48104 USA Phone: +1 248 816 6463 Email: chuck.lever@oracle.com Lever Expires August 7, 2017 [Page 17] RFC 3242 A Link-Layer Assisted ROHC RTP April 2002