FC-SP-Letter-ballot-comments

Document Sample
FC-SP-Letter-ballot-comments Powered By Docstoc
					FC-SP Revision 1.6 Comments
8/3/04 (04-535v0) Company-# Techn Physical Section/table/ Problem Description ical Page figure locator /Edito rial T x x.y.z State the problem here. E Cover Suggested solution Response Status Edit Status

Company-01 Brocade-01

Brocade-02 Brocade-03

E E

Cover Inside Cover

Brocade-04

E

All

Brocade-05

E

1

Propose a solution to the problem here. Cover The title in the project proposal was given as Make suggested change. Fibre Channel Avionics Environment – SCSI-3 Remote Direct Memory Access. That title should be on the document above the abbreviation. Cover Delete FAX number for Snively Make suggested change. Inside Cover It looks like there is a page boundary Make suggested change. problem here. I would have expected that the text beginning with the titled paragraph "INCITS Technical Report Series" belonged on the same page with the paragraphs on the next page. A page break should be placed above this paragraph. All The Scope and Foreword sections are Make suggested change. missing and should be supplied. In particular, most of Clause 1, first paragraph should be in the Foreword. All of the second paragraph of Clause 1 should be moved to the Foreword. 1 Remove first and second paragraph and Make suggested change. place them in the Foreword. Rewrite the third paragraph to include the official name and official abbreviation of the TR. Expand it to indicate that it is intended to provide RDMA support bi-directionally between two N_Ports in a constrained and carefully defined environment, typical of avionics applications. 1 The first sentence of the first paragraph Make suggested change. (which will be moved into the Foreword) says "pertinent to their use", but it is unclear whether their applies to the protocols and profiles or the fabric and loop. The sentence should be rewritten

Brocade-06

E

1

Page 1 of 5

FC-SP Revision 1.6 Comments
8/3/04 (04-535v0) Company-# Techn Physical Section/table/ Problem Description Suggested solution ical Page figure locator /Edito rial E 1 1 The fourth paragraph should replace "from Make suggested change. the standards described below, most notably from Fibre Channel...FC-AE-RDMA" with "from the normative references in clause 2." The "most notably" phrase does not belong here. It should have been mentioned that this is the mechanism used in the first paragraph of the rewritten clause 1 (presently the third in the clause). E T 2 2 2.2 2.1 Delete the empty clause 2 and delete the title for clause 2.1. Do we really want to use such obsolete documents as references? While simple, they do have some problems with inconsistency and error management. It should not be too hard to winnow out and prohibit those FCP-3 functions that are not used or useable in FC-AE-RDMA, like REC, SRP, and the retry identifiers. Documents this old may actually add cost by not having clean mappings to modern hardware implementations. The references should be indicated by their abbreviation wherever they are called upon. Specific examples will be given below, but the use of [2] or ANSI X3.269 is too far removed from the document to make it easy to sort out. If necessary, these document names should be placed in the abbreviations. This should be rewritten as: "A SCSI device containing application clients which originate device service and task management requests to be processed by a target SCSI device. (See SAM.) For FC-AERDMA, a device containing application clients that originate RDMA requests." Note that this definition has been removed from later revisions of SAM and replaced with "SCSI initiator", which of course would not be correct here. Make suggested change. Make suggested change. Response Status Edit Status

Brocade-07

Brocade-08 Brocade-09

Brocade-10

E

All

All

Make suggested change.

Brocade-11

T

3

3.1.2

Make suggested change.

Page 2 of 5

FC-SP Revision 1.6 Comments
8/3/04 (04-535v0) Company-# Techn Physical Section/table/ Problem Description ical Page figure locator /Edito rial E 3 3.1.3 This should be rewritten as: "a Target resident entity that implements a device model and executes SCSI commands sent by an application client. (See SAM.) For FC-AE-RDMA, Logical Unit is used to select among specific memory regions within a Target." E All All The bookmarks should include the section number in their format. E 3 3.1.4 "and a FCP_RSP_IU" should be changed to "and an FCP_RSP_IU". E 3 3.1.5 "[2]. RDMA has ... applications" should be replaced with ". RDMA restricts FCP to improve support for low-latency, real-time applications." It really doesn't enhance it, so much as it profiles it. T 3 3.1.6 This should be rewritten as: "A SCSI device that receives SCSI commands and directs such commands to one or more logical units for execution. (See SAM.) For FC-AE-RDMA, a device that accepts RDMA requests and performs the requested RDMA action. E 4 3.2 First paragraph. If they do not have their normal English meaning, what meaning do they have, or rather, where is their meaning defined? All such words must have either an inline definition where they are first used (if used rarely) or be included in the glossary with a reference to the source of the word (if used frequently). E 4 3.2 Fifth paragraph. The word shall is problematic in technical reports, which tend to be recommendations and profiles. Review the use of the word shall, and if possible eliminate it. Eighth paragraph should be rewritten to read: "The fields or control bits that are not applicable shall be set as required by the defining standard or technical report." Suggested solution Response Status Edit Status

Brocade-12

Make suggested change.

Brocade-13 Brocade-14 Brocade-15

Make suggested change. Make suggested change. Make suggested change.

Brocade-16

Make suggested change.

Brocade-17

Review text to be sure that all such words are defined.

Brocade-18

Make suggested change.

Brocade-19

E

4

3.2

Make suggested change.

Page 3 of 5

FC-SP Revision 1.6 Comments
8/3/04 (04-535v0) Company-# Techn Physical Section/table/ Problem Description ical Page figure locator /Edito rial E Many Many Page 4, 3.2, Tenth paragraph. Notes are descriptive and in general do not provide normative information. I suggest that the term in the tables be changed to "Additional Information" and that this paragraph be deleted. E Many Many Page 4, 3.2.1. These conventions are contrary to all other FC documents, which do not use the single quote to isolate such strings. I recommend using the FC conventions. E Many Many Remove hanging paragraphs. Example is 3.2. The text between 3.2 and 3.2.1 should be labeled 3.2.1 and titled "General Conventions". In other cases, words like "Introduction to ..." could be used. There are many other cases that also need to be corrected. E 5 3.3 The definition of FC-AE-RDMA should be changed to read: "Fibre Channel - Avionics Environment - Remote Direct Memory Access", this technical report" E 5 3.3 Remove "SCSI-3" and replace it with "(See SAM.) E 6 3.4 In the definition of Prohibited, the last sentence should be rewritten to: "Interoperability is not guaranteed if Prohibited features are required." T 7 4.2 Note 1 should be deleted, since there is no FC-AE-FCLP project defined. T All All The document does not appear to specify the required contents of the CDB field. All read/write, length, and displacement information is actually outside the CDB field. Data appears to be always transmitted by FCP_DATA sequences. There is one place that implies something different, but no substantive explanation is provided. (See 4.2.4.3.3.) In particular, is the information contained in the vendor-specific CDB metadata applied to the data transformed in the FCP_DATA IU, or is it the first 16 bytes of the vendor data. Suggested solution Response Status Edit Status

Brocade-20

Make suggested change.

Brocade-21

Make suggested change.

Brocade-22

Make suggested change.

Brocade-23

Make suggested change.

Brocade-24 Brocade-25

Make suggested change. Make suggested change.

Brocade-26 Brocade-27

Make suggested change. Resolve the question and explain in the text. If this is not a normative behavior, place an appropriate informative annex in the document to explain this.

Page 4 of 5

FC-SP Revision 1.6 Comments
8/3/04 (04-535v0) Company-# Techn Physical Section/table/ Problem Description ical Page figure locator /Edito rial T All All What happens if a real FCP device logs in? There is nothing to distinguish a real device from an FC-AE-RDMA device until the CDB is parsed, and at that point real initiators would expect the proper status and check conditions to be returned. What the FC-AERDMA device returns is not defined. T 8 4.2.3 Note that FCP-2 defined the parameter field in such a manner that a valid FC-AE-RDMA value may appear there (albeit with low probability). This is another of the error cases that must be considered as in Brocade-28. Suggested solution Response Status Edit Status

Brocade-28

Brocade-29

Brocade-30

T

8

4.2.3

Note 6 falls into this general area of "what happens if" explored in Brocade-28. I would suggest that this be administratively prohibited.

Brocade-31

T

9

4.2.4

Table 2 says that "Initiator Function" and "Target Function" shall both be 1. However, it looks like an asymmetrical definition where only one of the two devices was an initiator would also be a practical approach. What is actually intended here?

Examine this and related error conditions. Either prohibit such connections administratively as a requirement of the document or provide an error condition that will gracefully prevent interactions between FCP SCSI and FCP RDMA devices. Examine this and related error conditions. Either prohibit such connections administratively as a requirement of the document or provide an error condition that will gracefully prevent interactions between FCP SCSI and FCP RDMA devices. Examine this and related error conditions. Either prohibit such connections administratively as a requirement of the document or provide an error condition that will gracefully prevent interactions between FCP SCSI and FCP RDMA devices. Examine this question and explain it in the section describing the overall usage model of RDMA. This section is actually very weak at present and should be beefed up. I believe it is 4.2 and 4.2.1.

Page 5 of 5


				
DOCUMENT INFO
Shared By:
Tags: FC-SP, -Lett
Stats:
views:8
posted:11/28/2009
language:English
pages:5
Description: FC-SP-Letter-ballot-comments