idnits 2.17.1 draft-ietf-nfsv4-layout-types-09.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year (Using the creation date from RFC5661, updated by this document, for RFC5378 checks: 2005-10-21) -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (February 05, 2018) is 2266 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 5661 (Obsoleted by RFC 8881) == Outdated reference: A later version (-19) exists of draft-ietf-nfsv4-flex-files-13 Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 NFSv4 T. Haynes 3 Internet-Draft Primary Data 4 Updates: 5661 (if approved) February 05, 2018 5 Intended status: Standards Track 6 Expires: August 9, 2018 8 Requirements for pNFS Layout Types 9 draft-ietf-nfsv4-layout-types-09.txt 11 Abstract 13 This document defines the requirements which individual pNFS layout 14 types need to meet in order to work within the parallel NFS (pNFS) 15 framework as defined in RFC5661. In so doing, it aims to clearly 16 distinguish between requirements for pNFS as a whole and those 17 specifically directed to the pNFS File Layout. The lack of a clear 18 separation between the two set of requirements has been troublesome 19 for those specifying and evaluating new Layout Types. In this 20 regard, this document updates RFC5661. 22 Status of This Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on August 9, 2018. 39 Copyright Notice 41 Copyright (c) 2018 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 57 2. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 3 58 2.1. Use of the Terms "Data Server" and "Storage Device" . . . 5 59 2.2. Requirements Language . . . . . . . . . . . . . . . . . . 6 60 3. The Control Protocol . . . . . . . . . . . . . . . . . . . . 6 61 3.1. Control Protocol Requirements . . . . . . . . . . . . . . 8 62 3.2. Previously Undocumented Protocol Requirements . . . . . . 9 63 3.3. Editorial Requirements . . . . . . . . . . . . . . . . . 10 64 4. Specifications of Original Layout Types . . . . . . . . . . . 11 65 4.1. File Layout Type . . . . . . . . . . . . . . . . . . . . 11 66 4.2. Block Layout Type . . . . . . . . . . . . . . . . . . . . 12 67 4.3. Object Layout Type . . . . . . . . . . . . . . . . . . . 13 68 5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 69 6. Security Considerations . . . . . . . . . . . . . . . . . . . 15 70 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 15 71 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 15 72 8.1. Normative References . . . . . . . . . . . . . . . . . . 15 73 8.2. Informative References . . . . . . . . . . . . . . . . . 16 74 Appendix A. Acknowledgments . . . . . . . . . . . . . . . . . . 16 75 Appendix B. RFC Editor Notes . . . . . . . . . . . . . . . . . . 16 76 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 16 78 1. Introduction 80 The concept of layout type has a central role in the definition and 81 implementation of Parallel Network File System (pNFS) (see 82 [RFC5661]). Clients and servers implementing different layout types 83 behave differently in many ways while conforming to the overall pNFS 84 framework defined in [RFC5661] and this document. Layout types may 85 differ as to: 87 o The method used to do I/O operations directed to data storage 88 devices. 90 o The requirements for communication between the metadata server 91 (MDS) and the storage devices. 93 o The means used to ensure that I/O requests are only processed when 94 the client holds an appropriate layout. 96 o The format and interpretation of nominally opaque data fields in 97 pNFS-related NFSv4.x data structures. 99 Such matters are defined in a standards-track layout type 100 specification. Except for the files layout type, which was defined 101 in Section 13 of [RFC5661], existing layout types are defined in 102 their own standards-track documents and it is anticipated that new 103 layout types will be defined in similar documents. 105 The file layout type was defined in the Network File System (NFS) 106 version 4.1 protocol specification [RFC5661]. The block layout type 107 was defined in [RFC5663] while the object layout type was defined in 108 [RFC5664]. Subsequently, the SCSI layout type was defined in 109 [RFC8154]. 111 Some implementers have interpreted the text in Sections 12 ("Parallel 112 NFS (pNFS)") and 13 ("NFSv4.1 as a Storage Protocol in pNFS: the File 113 Layout Type") of [RFC5661] as both being applying only to the file 114 layout type. Because Section 13 was not covered in a separate 115 standards-track document such as those for both the block and object 116 layout types, there had been some confusion as to the 117 responsibilities of both the metadata server and the data servers 118 (DS) which were laid out in Section 12. 120 As a consequence, authors of new specifications (see [FlexFiles] and 121 [Lustre]) may struggle to meet the requirements to be a pNFS layout 122 type. This document gathers the requirements from all of the 123 original layout type standard documents and then specifies the 124 requirements placed on all layout types independent of the particular 125 type chosen. 127 2. Definitions 129 control communication requirements: are for a layout type the 130 details regarding information on layouts, stateids, file metadata, 131 and file data which must be communicated between the metadata 132 server and the storage devices. 134 control protocol: is the particular mechanism that an implementation 135 of a layout type would use to meet the control communication 136 requirement for that layout type. This need not be a protocol as 137 normally understood. In some cases the same protocol may be used 138 as a control protocol and storage protocol. 140 storage protocol: is the protocol used by clients to do I/O 141 operations to the storage device. Each layout type specifies the 142 set of storage protocols. 144 loose coupling: is when the control protocol is a storage protocol. 146 tight coupling: is an arrangement in which the control protocol is 147 one designed specifically for that purpose. It may be either a 148 proprietary protocol, adapted specifically to a a particular 149 metadata server, or one based on a standards-track document. 151 (file) data: is that part of the file system object which contains 152 the data to read or written. It is the contents of the object 153 rather than the attributes of the object. 155 data server (DS): is a pNFS server which provides the file's data 156 when the file system object is accessed over a file-based 157 protocol. Note that this usage differs from that in [RFC5661] 158 which applies the term in some cases even when other sorts of 159 protocols are being used. Depending on the layout, there might be 160 one or more data servers over which the data is striped. While 161 the metadata server is strictly accessed over the NFSv4.1 162 protocol, the data server could be accessed via any file access 163 protocol that meets the pNFS requirements. 165 See Section 2.1 for a comparison of this term and "data storage 166 device". 168 storage device: is the target to which clients may direct I/O 169 requests when they hold an appropriate layout. Note that each 170 data server is a storage device but that some storage device are 171 not data servers. See Section 2.1 for further discussion. 173 fencing: is the process by which the metadata server prevents the 174 storage devices from processing I/O from a specific client to a 175 specific file. 177 layout: is the information a client uses to access file data on a 178 storage device. This information will include specification of 179 the protocol (layout type) and the identity of the storage devices 180 to be used. 182 The bulk of the contents of the layout are defined in [RFC5661] 183 as nominally opaque, but individual layout types are responsible 184 for specifying the format of the layout data. 186 layout iomode: is a grant of either read or read/write I/O to the 187 client. 189 layout stateid: is a 128-bit quantity returned by a server that 190 uniquely defines the layout state provided by the server for a 191 specific layout that describes a layout type and file (see 192 Section 12.5.2 of [RFC5661]). Further, Section 12.5.3 describes 193 differences in handling between layout stateids and other stateid 194 types. 196 layout type: is a specification of both the storage protocol used to 197 access the data and the aggregation scheme used to lay out the 198 file data on the underlying storage devices. 200 recalling a layout: is a graceful recall, via a callback, of a 201 specific layout by the metadata server to the client. Graceful 202 here means that the client would have the opportunity to flush any 203 writes, etc., before returning the layout to the metadata server. 205 revoking a layout: is an invalidation of a specific layout by the 206 metadata server. Once revocation occurs, the metadata server will 207 not accept as valid any reference to the revoked layout and a 208 storage device will not accept any client access based on the 209 layout. 211 (file) metadata: is that part of the file system object that 212 contains various descriptive data relevant to the file object, as 213 opposed to the file data itself. This could include the time of 214 last modification, access time, end-of-file (EOF) position, etc. 216 metadata server (MDS): is the pNFS server which provides metadata 217 information for a file system object. It also is responsible for 218 generating, recalling, and revoking layouts for file system 219 objects, for performing directory operations, and for performing I 220 /O operations to regular files when the clients direct these to 221 the metadata server itself. 223 stateid: is a 128-bit quantity returned by a server that uniquely 224 defines the set of locking-related state provided by the server. 225 Stateids may designate state related to open files, to byte-range 226 locks, to delegations, or to layouts. 228 2.1. Use of the Terms "Data Server" and "Storage Device" 230 In [RFC5661], these two terms of "Data Server" and "Storage Device" 231 are used somewhat inconsistently: 233 o In chapter 12, where pNFS in general is discussed, the term 234 "storage device" is used. 236 o In chapter 13, where the file layout type is discussed, the term 237 "data server" is used. 239 o In other chapters, the term "data server" is used, even in 240 contexts where the storage access type is not NFSv4.1 or any other 241 file access protocol. 243 As this document deals with pNFS in general, it uses the more generic 244 term "storage device" in preference to "data server". The term "data 245 server" is used only in contexts in which a file server is used as a 246 storage device. Note that every data server is a storage device but 247 storage devices which use protocols which are not file access 248 protocols (such as NFS) are not data servers. 250 Since a given storage device may support multiple layout types, a 251 given device can potentially act as a data server for some set of 252 storage protocols while simultaneously acting as a storage device for 253 others. 255 2.2. Requirements Language 257 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 258 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 259 document are to be interpreted as described in [RFC2119]. 261 This document differs from most standards-track documents in that it 262 specifies requirements for those defining future layout types rather 263 than defining the requirements for implementations directly. This 264 document makes clear whether: 266 (1) any particular requirement applies to implementations. 268 (2) any particular requirement applies to those defining layout 269 types. 271 (3) the requirement is a general requirement which implementations 272 need to conform to, with the specific means left to layout type 273 definitions type to specify. 275 3. The Control Protocol 277 A layout type has to meet the requirements that apply to the 278 interaction between the metadata server and the storage device such 279 that they present to the client a consistent view of stored data and 280 lock state (Section 12.2.6 of [RFC5661]). Particular implementations 281 may satisfy these requirements in any manner they choose and the 282 mechanism chosen need not be described as a protocol. Specifications 283 defining layout types need to clearly show how implementations can 284 meet the requirements discussed below, especially with respect to 285 those that have security implications. In addition, such 286 specifications may find it necessary to impose requirements on 287 implementations of the layout type to ensure appropriate 288 interoperability. 290 In some cases, there may be no control protocol other than the 291 storage protocol. This is often described as using a "loose 292 coupling" model. In such cases, the assumption is that the metadata 293 server, storage devices, and client may be changed independently and 294 that the implementation requirements in the layout type specification 295 need to ensure this degree of interoperability. This model is used 296 in the block and object layout type specification. 298 In other cases, it is assumed that there will be a purpose-built 299 control protocol which may be different for different implementations 300 of the metadata server and data server. The assumption here is that 301 the metadata server and data servers are designed and implemented as 302 a unit and interoperability needs to be assured between clients and 303 metadata-data server pairs, developed independently. This is the 304 model used for the files layout. 306 Another possibility is for the definition of a control protocol to be 307 specified in a standards-track document. There are two subcases to 308 consider: 310 o A new layout type includes a definition of a particular control 311 protocol whose use is obligatory for metadata servers and storage 312 devices implementing the layout type. In this case the 313 interoperability model is similar to the first case above and the 314 defining document should assure interoperability among metadata 315 servers, storage devices, and clients developed independently. 317 o A control protocol is defined in a standards-track document which 318 meets the control protocol requirements for one of the existing 319 layout types. In this case, the new document's job is to assure 320 interoperability between metadata servers and storage devices 321 developed separately. The existing definition document for the 322 selected layout type retains the function of assuring 323 interoperability between clients and a given collection of 324 metadata servers and storage devices. In this context, 325 implementations that implement the new protocol are treated in the 326 same way as those that use an internal control protocol or a 327 functional equivalent. 329 An example of this last case is the SCSI layout type [RFC8154], which 330 extends the block layout type. The block layout type had a 331 requirement for fencing of clients, but did not present a way for the 332 control protocol (in this case the SCSI storage protocol) to fence 333 the client. The SCSI layout type remedies that in [RFC8154] and in 334 effect has a tightly coupled model. 336 3.1. Control Protocol Requirements 338 The requirements of interactions between the metadata server and the 339 storage devices are: 341 (1) The metadata server MUST be able to service the client's I/O 342 requests if the client decides to make such requests to the 343 metadata server instead of to the storage device. The metadata 344 server must be able to retrieve the data from the constituent 345 storage devices and present it back to the client. A corollary 346 to this is that even though the metadata server has successfully 347 given the client a layout, the client MAY still send I/O 348 requests to the metadata server. 350 (2) The metadata server MUST be able to restrict access to a file on 351 the storage devices when it revokes a layout. The metadata 352 server typically would revoke a layout whenever a client fails 353 to respond to a recall or a client's lease is expired due to 354 non-renewal. It might also revoke the layout as a means of 355 enforcing a change in locking state or access permissions that 356 the storage device cannot directly enforce. 358 Effective revocation may require client co-operation in using a 359 particular stateid (files layout) or principal (e,g., flexible 360 files layout) when performing I/O. 362 In contrast, there is no requirement to restrict access to a 363 file on the storage devices when a layout is recalled. It is 364 only after the metadata server determines that the client is not 365 gracefully returning the layout and starts the revocation that 366 this requirement is enforced. 368 (3) A pNFS implementation MUST NOT allow the violation of NFSv4.1's 369 access controls: ACLs and file open modes. Section 12.9 of 370 [RFC5661] specifically lays this burden on the combination of 371 clients, storage devices, and the metadata server. However the 372 specification of the individual layout type might create 373 requirements as to how this is to be done. This may include a 374 possible requirement for the metadata server to update the 375 storage device so that it can enforce security. 377 The file layout requires the storage device to enforce access 378 whereas the flex file layout requires both the storage device 379 and the client to enforce security. 381 (4) Interactions between locking and I/O operations MUST obey 382 existing semantic restrictions. In particular, if an I/O 383 operation would be invalid when directed at the metadata server, 384 it is not to be allowed when performed on the storage device. 386 For the block and SCSI layout, as the storage device is not able 387 to reject the I/O operation, the client is responsible for 388 enforcing this requirement. 390 (5) Any disagreement between the metadata server and the data server 391 as to the value of attributes such as modify time, the change 392 attribute, and the EOF position MUST be of limited duration with 393 clear means of resolution of any discrepancies being provided. 394 Note that 396 (a) Discrepancies need not be resolved unless any client has 397 accessed the file in question via the metadata server, 398 typically by performing a GETATTR. 400 (b) A particular storage device might be striped such it has no 401 information regarding the EOF position. 403 (c) Both clock skew and network delay can lead to the metadata 404 server and the storage device having different values of 405 the time attributes. As long as those differences can be 406 accounted for in what is presented to the client in a 407 GETATTR, then no violation results. 409 (d) A LAYOUTCOMMIT requires that changes in attributes 410 resulting from operations on the storage device need to be 411 reflected in the metadata server by the completion of the 412 operation. 414 These requirements may be satisfied in different ways by different 415 layout types. As an example, while the file layout type uses the 416 stateid to fence off the client, there is no requirement that other 417 layout types use this stateid approach. 419 Each new standards-track document for a layout types MUST address how 420 the client, metadata server, and storage devices are to interact to 421 meet these requirements. 423 3.2. Previously Undocumented Protocol Requirements 425 While not explicitly stated as requirements in Section 12 of 426 [RFC5661], the existing layout types do have more requirements that 427 they need to enforce. 429 The client has these obligations when making I/O requests to the 430 storage devices: 432 (1) Clients MUST NOT perform I/O to the storage device if they do 433 not have layouts for the files in question. 435 (2) Clients MUST NOT perform I/O operations outside of the specified 436 ranges in the layout segment. 438 (3) Clients MUST NOT perform I/O operations which would be 439 inconsistent with the iomode specified in the layout segments it 440 holds. 442 Under the file layout type, the storage devices are able to reject 443 any request made not conforming to these requirements. This may not 444 be possible for other known layout types, which puts the burden of 445 enforcing such violations solely on the client. For these layout 446 types: 448 (1) The metadata server MAY use fencing operations to the storage 449 devices to enforce layout revocation against the client. 451 (2) The metadata server MUST allow the clients to perform data I/O 452 against it, even if it has already granted the client a layout. 453 A layout type might discourage such I/O, but it can not forbid 454 it. 456 (3) The metadata server MUST be able to do storage allocation, 457 whether that is to create, delete, extend, or truncate files. 459 The means to address these requirements will vary with the layout 460 type. A control protocol will be used to effect these, whether a 461 purpose-built one, one identical to the storage protocol, or a new 462 standards-track control protocol. 464 3.3. Editorial Requirements 466 This section discusses how the protocol requirements discussed above 467 need to be addressed in documents specifying a new layout type. 468 Depending on the interoperability model for the layout type in 469 question, this may involve the imposition of layout-type-specific 470 requirements that ensure appropriate interoperability of pNFS 471 components which are developed separately. 473 The specification of the layout type needs to make clear how the 474 client, metadata server, and storage device act together to meet the 475 protocol requirements discussed previously. If the document does not 476 impose implementation requirements sufficient to ensure that these 477 semantic requirements are met, it is not appropriate for the working 478 group to allow the document to move forward. 480 Some examples include: 482 o If the metadata server does not have a means to invalidate a 483 stateid issued to the storage device to keep a particular client 484 from accessing a specific file, then the layout type specification 485 has to document how the metadata server is going to fence the 486 client from access to the file on that storage device. 488 o If the metadata server implements mandatory byte-range locking 489 when accessed directly by the client, then the layout type 490 specification must require that this also be done when data is 491 read or written using the designated storage protocol. 493 4. Specifications of Original Layout Types 495 This section updates Section 12 of [RFC5661], which enumerates the 496 requirements of pNFS layout type specifications. It is not normative 497 with regards to the file layout type presented in Section 13 of 498 [RFC5661], the block layout type [RFC5663], and the object layout 499 type [RFC5664]. These are discussed here only to illuminate the 500 updates made to Section 12 of [RFC5661]. 502 4.1. File Layout Type 504 Because the storage protocol is a subset of NFSv4.1, the semantics of 505 the file layout type comes closest to the semantics of NFSv4.1 in the 506 absence of pNFS. In particular, the stateid and principal used for I 507 /O MUST have the same effect and be subject to the same validation on 508 a data server as it would have if the I/O were being performed on the 509 metadata server itself. The same set of validations are applied 510 whether pNFS is in effect or not. 512 And while for most implementations the storage devices can do the 513 following validations: 515 (1) client holds a valid layout, 517 (2) client I/O matches the layout iomode, and, 519 (3) client does not go out of the byte ranges, 521 these are each presented as a "SHOULD" and not a "MUST". Actually, 522 the first point is presented in [RFC5661] as both: 524 "MUST": in Section 13.6 525 "As described in Section 12.5.1, a client MUST NOT send an I/O to 526 a data server for which it does not hold a valid layout; the data 527 server MUST reject such an I/O." 529 "SHOULD": in Section 13.8 531 "The iomode need not be checked by the data servers when clients 532 perform I/O. However, the data servers SHOULD still validate that 533 the client holds a valid layout and return an error if the client 534 does not." 536 It should be noted that it is just these layout specific checks that 537 are optional, not the normal file access semantics. The storage 538 devices MUST make all of the required access checks on each READ or 539 WRITE I/O as determined by the NFSv4.1 protocol. If the metadata 540 server would deny a READ or WRITE operation on a file due to its ACL, 541 mode attribute, open access mode, open deny mode, mandatory byte- 542 range lock state, or any other attributes and state, the storage 543 device MUST also deny the READ or WRITE operation. Also while the 544 NFSv4.1 protocol does not mandate export access checks based on the 545 client's IP address, if the metadata server implements such a policy, 546 then that counts as such state as outlined above. 548 The data filehandle provided by the PUTFH operation to the data 549 server provides sufficient context to enable the data server to 550 ensure that for the subsequent READ or WRITE operation in the 551 compound, that the client has a valid layout for the I/O being 552 performed. 554 Finally, the data server can check the stateid presented in the READ 555 or WRITE operation to see if that stateid has been rejected by the 556 metadata server in order to cause the I/O to be fenced. Whilst it 557 might just be the open owner or lock owner on that client being 558 fenced, the client should take the NFS4ERR_BAD_STATEID error code to 559 mean it has been fenced from the file and contact the metadata 560 server. 562 4.2. Block Layout Type 564 With the block layout type, the storage devices are generally not 565 able to enforce file-based security. Typically, storage area network 566 (SAN) disk arrays and SAN protocols provide coarse-grained access 567 control mechanisms (e.g., Logical Unit Number (LUN) mapping and/or 568 masking), with a target granularity of disks rather than individual 569 blocks and a source granularity of individual hosts rather than of 570 users or owners. Access to block storage is logically at a lower 571 layer of the I/O stack than NFSv4. Since NFSv4 security is not 572 directly applicable to protocols that access such storage directly, 573 Section 2.1 [RFC5663] specifies that: 575 "in environments where pNFS clients cannot be trusted to enforce 576 such policies, pNFS block layout types SHOULD NOT be used." 578 Due to these granularity issues, the security burden has been shifted 579 from the storage devices to the client. Those deploying 580 implementations of this layout type need to be sure that the client 581 implementation can be trusted This is not a new sort of requirement 582 in the context of SAN protocols. In such environments, the client is 583 expected to provide block-based protection. 585 This shift of the burden also extends to locks and layouts. The 586 storage devices are not able to enforce any of these and the burden 587 is pushed to the client to make the appropriate checks before sending 588 I/O to the storage devices. For example, the server may use a layout 589 iomode only allowing reading to enforce a mandatory read-only lock, 590 In such cases, the client has to support that use by not sending 591 WRITEs to the storage devices. The fundamental issue here is that 592 the storage device is treated by this layout type in the same fashion 593 as a local disk device. Once the client has access to the storage 594 device, it is able to perform both READ and WRITE I/O to the entire 595 storage device. The byte ranges in the layout, any locks, the layout 596 iomode, etc, can only be enforced by the client. Therefore, the 597 client is required to provide that enforcement. 599 In the context of fencing off of the client upon revocation of a 600 layout, these limitations come into play again, i.e., the granularity 601 of the fencing can only be at the host/logical-unit level. Thus, if 602 one of a client's layouts is revoked by the server, it will 603 effectively revoke all of the client's layouts for files located on 604 the storage units comprising the logical volume. This may extend to 605 the client's layouts for files in other file systems. Clients need 606 to be prepared for such revocations and reacquire layouts as needed. 608 4.3. Object Layout Type 610 With the object layout type, security checks occur during the 611 allocation of the layout. The client will typically ask for layouts 612 covering all of the file and may do so for either READ or READ/WRITE. 613 This enables it to do subsequent I/O operations without the need to 614 obtain layouts for specific byte ranges. At that time, the metadata 615 server should verify permissions against the layout iomode, the file 616 mode bits or ACLs, etc. As the client may be acting for multiple 617 local users, it MUST authenticate and authorize the user by issuing 618 respective OPEN and ACCESS calls to the metadata server, similar to 619 having NFSv4 data delegations. 621 Upon successful authorization, the client receives within the layout 622 a set of object capabilities allowing it I/O access to the specified 623 objects corresponding to the requested iomode. These capabilities 624 are used to enforce access control and locking semantics at the 625 storage devices. Whenever one of the following occur on the metadata 626 server: 628 o the permissions on the object change, 630 o a conflicting mandatory byte-range lock is granted, or 632 o a layout is revoked and reassigned to another client, 634 then the metadata server MUST change the capability version attribute 635 on all objects comprising the file to in order to invalidate any 636 outstanding capabilities before committing to one of these changes. 638 When the metadata server wishes to fence off a client to a particular 639 object, then it can use the above approach to invalidate the 640 capability attribute on the given object. The client can be informed 641 via the storage device that the capability has been rejected and is 642 allowed to fetch a refreshed set of capabilities, i.e., re-acquire 643 the layout. 645 5. Summary 647 In the three original layout types, the burden of enforcing the 648 security of NFSv4.1 can fall to either the storage devices (files), 649 the client (blocks), or the metadata server (objects). Such choices 650 are conditioned by the native capabilities of the storage devices - 651 if a control protocol can be implemented, then the burden can be 652 shifted primarily to the storage devices. 654 In the context of this document, we treat the control protocol as a 655 set of requirements. And as new layout types are published, the 656 defining documents MUST address: 658 (1) The fencing of clients after a layout is revoked. 660 (2) The security implications of the native capabilities of the 661 storage devices with respect to the requirements of the NFSv4.1 662 security model. 664 In addition, these defining documents need to make clear how other 665 semantic requirements of NFSv4.1 (e.g., locking) are met in the 666 context of the proposed layout type. 668 6. Security Considerations 670 This section does not deal directly with security considerations for 671 existing or new layout types. Instead, it provides a general 672 framework for understating security-related issues within the pNFS 673 framework. Specific security considerations will be addressed in the 674 Security Considerations sections of documents specifying layout 675 types. For example, in Section 5 of [RFC5663], the lack of finer- 676 than-physical disk access control necessitates that the client is 677 delegated the responsibility to enforce the access provided to them 678 in the layout extent which they were granted by the metadata server. 680 The layout type specification must ensure that only data accesses 681 consistent with the NFSV4.1 security model are allowed. It may do 682 this directly, by providing that appropriate checks be performed at 683 the time each access is performed. It may do it indirectly by 684 allowing the client or the storage device to be responsible for 685 making the appropriate checks. In the latter case, I/O access writes 686 are reflected in layouts and the layout type must provide a way to 687 prevent inappropriate access due to permissions changes between the 688 time a layout is granted and the time the access is performed. 690 The metadata server MUST be able to fence off a client's access to 691 the data file on a storage device. When it revokes the layout, the 692 client's access MUST be terminated at the storage devices. The 693 client has a subsequent opportunity to re-acquire the layout and 694 perform the security check in the context of the newly current access 695 permissions. 697 7. IANA Considerations 699 This document has no actions for IANA. 701 8. References 703 8.1. Normative References 705 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 706 Requirement Levels", March 1997. 708 [RFC5661] Shepler, S., Eisler, M., and D. Noveck, "Network File 709 System (NFS) Version 4 Minor Version 1 Protocol", RFC 710 5661, January 2010. 712 [RFC5663] Black, D., Fridella, S., and J. Glasgow, "pNFS Block/ 713 Volume Layout", RFC 5663, January 2010. 715 [RFC5664] Halevy, B., Welch, B., and J. Zelenka, "Object-Based 716 Parallel NFS (pNFS) Operations", RFC 5664, January 2010. 718 [RFC8154] Hellwig, C., "Parallel NFS (pNFS) Small Computer System 719 Interface (SCSI) Layout", RFC 8154, DOI 10.17487/RFC8154, 720 May 2017, . 722 8.2. Informative References 724 [FlexFiles] 725 Halevy, B. and T. Haynes, "Parallel NFS (pNFS) Flexible 726 File Layout", draft-ietf-nfsv4-flex-files-13 (Work In 727 Progress), August 2017. 729 [Lustre] Faibish, S. and P. Tao, "Parallel NFS (pNFS) Lustre Layout 730 Operations", draft-faibish-nfsv4-pnfs-lustre-layout-07 731 (Work In Progress), April 2014. 733 Appendix A. Acknowledgments 735 Dave Noveck provided an early review that sharpened the clarity of 736 the definitions. He also provided a more comprehensive review of the 737 document. 739 Both Chuck Lever and Christoph Helwig provided insightful comments 740 during the WGLC. 742 Appendix B. RFC Editor Notes 744 [RFC Editor: please remove this section prior to publishing this 745 document as an RFC] 747 [RFC Editor: prior to publishing this document as an RFC, please 748 replace all occurrences of RFCTBD10 with RFCxxxx where xxxx is the 749 RFC number of this document] 751 Author's Address 753 Thomas Haynes 754 Primary Data, Inc. 755 4300 El Camino Real Ste 100 756 Los Altos, CA 94022 757 USA 759 Phone: +1 408 215 1519 760 Email: thomas.haynes@primarydata.com