pNFS Lustre Layout Operations
draft-faibish-nfsv4-pnfs-lustre-layout-00
The information below is for an old version of the document.
| Document | Type |
This is an older version of an Internet-Draft whose latest revision state is "Expired".
|
|
|---|---|---|---|
| Authors | Sorin Faibish , Peng Tao | ||
| Last updated | 2012-10-05 | ||
| RFC stream | (None) | ||
| Formats | |||
| Additional resources | |||
| Stream | Stream state | (No stream defined) | |
| Consensus boilerplate | Unknown | ||
| RFC Editor Note | (None) | ||
| IESG | IESG state | I-D Exists | |
| Telechat date | (None) | ||
| Responsible AD | (None) | ||
| Send notices to | (None) |
draft-faibish-nfsv4-pnfs-lustre-layout-00
NFSv4 Working Group S. Faibish
Internet-Draft Peng Tao
Intended status: draft EMC Corporation
Expires: April 5, 2013 October 5, 2012
pNFS Lustre Layout Operations
draft-faibish-nfsv4-pnfs-lustre-layout-00
Status of this Memo
This Internet-Draft is submitted to IETF in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
This Internet-Draft will expire on April 5, 2013.
Copyright Notice
Copyright (c) 2011 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with
respect to this document. Code Components extracted from this
document must include Simplified BSD License text as described in
Faibish et al. Expires April 5, 2013 [Page 1]
Internet-Draft pNFS Lustre Layout Operations October 2012
Section 4.e of the Trust Legal Provisions and are provided without
warranty as described in the Simplified BSD License.
Abstract
Parallel NFS (pNFS) extends Network File System version 4.1 (NFSv4.1)
to allow clients to directly access file data on the storage used by
the NFSv4.1 server. This ability to bypass the server for data access
can increase both performance and parallelism, but requires
additional client functionality for data access, some of which is
dependent on the class of storage used, a.k.a. the Layout Type. The
main pNFS operations and data types in NFSv4 Minor version 1 specify
a layout-type-independent layer; layout-type-specific information is
conveyed using opaque data structures whose internal structure is
further defined by the particular layout type specification. This
document specifies the NFSv4.1 Lustre pNFS Layout Type as a companion
to the main NFSv4 Minor version 1 specification.
Table of Contents
1. Introduction...................................................3
2. Conventions used in this document..............................4
3. XDR Description of the Lustre-Based Layout Protocol............4
3.1. Code Components Licensing Notice..........................5
4. Basic Data Type Definitions....................................6
4.1. pnfs_lov_magic............................................6
4.2. pnfs_los_object_cred4.....................................7
4.3. data stripping algorithms.................................8
5. Object Storage Server Addressing and Discovery.................8
5.1. pnfs_los_targetid_type4...................................9
5.2. pnfs_los_deviceaddr4......................................9
6. Lustre-Based Layout...........................................10
6.1. pnfs_lov_mds_md..........................................10
6.2. pnfs_los_layout4.........................................12
6.3. Data Mapping Schemes.....................................13
6.4. RAID Algorithms..........................................14
7. Lustre-Based Creation Layout Hint.............................15
7.1. pnfs_los_layouthint4.....................................15
8. References....................................................18
8.1. Normative References.....................................18
Authors' Addresses...............................................19
Faibish et al. Expires April 5, 2013 [Page 2]
Internet-Draft pNFS Lustre Layout Operations October 2012
1. Introduction
Figure 1 shows the overall architecture of a Parallel NFS (pNFS)
system:
+-----------+
|+-----------+ +-----------+
||+-----------+ | |
||| | NFSv4.1 + pNFS | |
+|| Clients |<------------------------------>| MDS |
+| | | |
+-----------+ | |
||| +-----------+
||| |
||| |
||| Storage +-----------+ |
||| Protocol |+-----------+ |
||+----------------||+-----------+ Control |
|+-----------------||| | Protocol |
+------------------+|| Storage |------------+
+| Devices |
+-----------+
Figure 1 pNFS Architecture
In this document, "storage device" is used as a general term for a
data server and/or storage server for all pNFS layouts. The MetaData
Server (MDS) is the NFSv4.1 server that provides pNFS layouts to
clients and handles operations on file metadata (e.g., names,
attributes).
In pNFS, the file server returns typed layout structures that
describe where file data is located. There are different layouts for
different storage systems and methods of arranging data on storage
devices. This document describes the layouts used with Lustre storage
devices (OSSs) that are accessed according to the Lustre storage
protocol ([1]).
Lustre is an object-based file system. It is composed of three
components: Metadata servers (MDSs), object storage servers (OSSs),
and clients.
Lustre uses block devices (SCSI LUNs) for file data (called OST) and
metadata storages (MDT) and each block device can be managed by only
one Lustre service (OSS server). The total data capacity of the
Lustre filesystem is the sum of all individual OST capacities (such
Faibish et al. Expires April 5, 2013 [Page 3]
Internet-Draft pNFS Lustre Layout Operations October 2012
as SCSI LUNs). Lustre clients access and concurrently use data
through the standard POSIX I/O system calls.
An MDS (metadata server) provides metadata services. One MDS manages
one metadata target (MDT). Each MDT (SCSI LUN) stores file metadata,
such as file names, directory structures, and access permissions. An
OSS (object storage server) exposes block devices and serves data.
Each OSS manages one or more object storage targets (OSTs), and OSTs
store file data "objects".
The Lustre protocol specifies several operations on objects,
including OPEN, READ, WRITE, GET ATTRIBUTES, SET ATTRIBUTES, CREATE,
and DELETE. However, using the Lustre layout the client only uses the
OPEN, READ, WRITE and GET ATTRIBUTES commands. The other commands are
only used by the pNFS server.
A Lustre file object's layout information is defined in the extended
attribute (EA) of the inode. Essentially, EA describes the mapping
between file object id and its corresponding OSTs. Essentially, EA
describes the mapping between file object identifier and its
corresponding OSTs. This information is also known as striping. A
Lustre-based layout for pNFS includes object identifiers,
capabilities that allow clients to READ or WRITE those objects, and
various parameters that control how file data is striped across OSTs.
This document specifies the layout protocol and operations using as
data and control protocols the Lustre protocol ([1]).
2. Conventions used in this document
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 [6].
3. XDR Description of the Lustre-Based Layout Protocol
This document contains the external data representation (XDR [2])
description of the NFSv4.1 objects layout protocol. The XDR
description is embedded in this document in a way that makes it
simple for the reader to extract into a ready-to-compile form. The
reader can feed this document into the following shell script to
produce the machine readable XDR description of the NFSv4.1 Lustre
layout protocol:
#!/bin/sh
grep '^ *///' $* | sed 's?^ */// ??' | sed 's?^ *///$??'
Faibish et al. Expires April 5, 2013 [Page 4]
Internet-Draft pNFS Lustre Layout Operations October 2012
That is, if the above script is stored in a file called "extract.sh",
and this document is in a file called "spec.txt", then the reader can
do:
sh extract.sh < spec.txt > pnfs_lustre_prot.x
The effect of the script is to remove leading white space from each
line, plus a sentinel sequence of "///".
The embedded XDR file header follows. Subsequent XDR descriptions,
with the sentinel sequence are embedded throughout the document.
Note that the XDR code contained in this document depends on types
from the NFSv4.1 nfs4_prot.x file ([3]). This includes both nfs types
that end with a 4, such as offset4, length4, etc., as well as more
generic types such as uint32_t and uint64_t.
3.1. Code Components Licensing Notice
The XDR description, marked with lines beginning with the sequence
"///", as well as scripts for extracting the XDR description are Code
Components as described in Section 4 of "Legal Provisions Relating to
IETF Documents" [4]. These Code Components are licensed according to
the terms of Section 4 of "Legal Provisions Relating to IETF
Documents".
/// /*
/// * Copyright (c) 2010 IETF Trust and the persons identified
/// * as authors of the code. All rights reserved.
/// *
/// * Redistribution and use in source and binary forms, with
/// * or without modification, are permitted provided that the
/// * following conditions are met:
/// *
/// * o Redistributions of source code must retain the above
/// * copyright notice, this list of conditions and the
/// * following disclaimer.
/// *
/// * o Redistributions in binary form must reproduce the above
/// * copyright notice, this list of conditions and the
/// * following disclaimer in the documentation and/or other
/// * materials provided with the distribution.
/// *
/// * o Neither the name of Internet Society, IETF or IETF
/// * Trust, nor the names of specific contributors, may be
/// * used to endorse or promote products derived from this
/// * software without specific prior written permission.
Faibish et al. Expires April 5, 2013 [Page 5]
Internet-Draft pNFS Lustre Layout Operations October 2012
/// *
/// * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS
/// * AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED
/// * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
/// * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
/// * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
/// * EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
/// * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
/// * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
/// * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
/// * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
/// * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
/// * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
/// * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
/// * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
/// * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
/// *
/// * Please reproduce this note if possible.
/// */
///
/// /*
/// * pnfs_lustre_prot.x
/// */
///
/// %#include <nfs4_prot.x>
///
4. Basic Data Type Definitions
The following sections define basic data types and constants used by
the Lustre Layout protocol.
4.1. pnfs_lov_magic
Lustre uses two magic numbers to identify different lov_mds_md
versions.
/// enum pnfs_lov_magic {
/// LOV_MAGIC_V1 = 0x0BD10BD0 /* to identify lov_mds_md_v1 */
/// LOV_MAGIC_V3 = 0x0BD30BD0 /* to identify lov_mds_md_v3 */
/// };
pnfs_lov_magic is used to indicate the Lustre protocol MDS metadata
version.
Faibish et al. Expires April 5, 2013 [Page 6]
Internet-Draft pNFS Lustre Layout Operations October 2012
At this time, the Lustre protocol is using LOV_MAGIC_V1/3 to mark
different version of lov_mds_md. If OST pooling is used the server
will return LOV_MAGIC_V3. If OST pooling is not configured, the MDS
server SHOULD return LOV_MAGIC_V1. So the versioning is used just
for feature matching. The latest Lustre protocol V3 matches the
relevant data structures, APIs, protocols, and algorithms involved
for the Lustre version 1.6 source code base.
Therefore, the Lustre protocol version is explicitly called out in
the information returned in the layout. (The format value is
0x0BD10BD0 for version V1 capability. However, it seems most robust
to call out the magic number explicitly.)
4.2. pnfs_los_object_cred4
/// enum pnfs_los_cap_key_sec4 {
/// PNFS_OSS_CAP_KEY_SEC_NONE = 0,
/// PNFS_OSS_CAP_KEY_SEC_SSV = 1
/// };
///
/// struct pnfs_los_object_cred4 {
/// pnfs_los_objid4 lc_object_id;
/// pnfs_los_cap_key_sec4 lc_cap_key_sec;
/// opaque lc_capability_key<>;
/// opaque lc_capability<>;
/// };
///
Lustre ptlrpc supports gss authentication. So pnfs_los_object_cred4
is put inside pnfs_los_layout4 so that if network requires security,
credentials can be passed around.
The pnfs_los_object_cred4 structure is used to identify each
component comprising the file. The "lc_object_id" identifies the
component object, the "lc_capability_key" provide the OSS
security credentials needed to access that object. The
"lc_cap_key_sec" value denotes the method used to secure the
lc_capability_key.
To comply with the Lustre security requirements, the capability key
SHOULD be transferred securely to prevent eavesdropping. Therefore, a
client SHOULD either issue the LAYOUTGET or GETDEVICEINFO operations
via RPCSEC_GSS with the privacy service or previously establish a
secret state verifier (SSV) for the sessions via the NFSv4.1 SET_SSV
operation. The pnfs_los_cap_key_sec4 type is used to identify the
method used by the server to secure the capability key.
Faibish et al. Expires April 5, 2013 [Page 7]
Internet-Draft pNFS Lustre Layout Operations October 2012
o PNFS_OSS_CAP_KEY_SEC_NONE denotes that the lc_capability_key is
not encrypted, in which case the client SHOULD issue the LAYOUTGET
or GETDEVICEINFO operations with RPCSEC_GSS with the privacy
service or the NFSv4.1 transport should be secured by using
methods that are external to NFSv4.1 like the use of IPsec [5] for
transporting the NFSV4.1 protocol.
o PNFS_OSS_CAP_KEY_SEC_SSV denotes that the lc_capability_key
contents are encrypted using the SSV GSS context and the
capability key as inputs to the GSS_Wrap() function (see GSS-API
[7]) with the conf_req_flag set to TRUE. The client MUST use the
secret SSV key as part of the client's GSS context to decrypt the
capability key using the value of the lc_capability_key field as
the input_message to the GSS_unwrap() function. Note that to
prevent eavesdropping of the SSV key, the client SHOULD issue
SET_SSV via RPCSEC_GSS with the privacy service.
The actual method chosen depends on whether the client established a
SSV key with the server and whether it issued the operation with the
RPCSEC_GSS privacy method. Naturally, if the client did not
establish an SSV key via SET_SSV, the server MUST use the
PNFS_OSS_CAP_KEY_SEC_NONE method. Otherwise, if the operation was
not issued with the RPCSEC_GSS privacy method, the server SHOULD
secure the lc_capability_key with the PNFS_OSS_CAP_KEY_SEC_SSV
method. The server MAY use the PNFS_OSS_CAP_KEY_SEC_SSV method also
when the operation was issued with the RPCSEC_GSS privacy method.
4.3. data stripping algorithms
Currently only RAID0 is supported but Lustre defines RAID1 as well.
/// const LOV_PATTERN_RAID0 = 0x001 /* stripes are used round-robin */
/// const LOV_PATTERN_RAID1 = 0x002 /* stripes are mirrors of each
other */
5. Object Storage Server Addressing and Discovery
Data operations to an OSS require the client to know the "address" of
each OSS's root object. The OSS (object storage server) exposes block
devices and serves data. Correspondingly, OSC (object storage client)
is client of the services. Each OSS manages one or more object storage
targets (OSTs), and OSTs store file data objects. Because these
representations are local, GETDEVICEINFO must return information that
can be used by the client to select the correct local representation.
Faibish et al. Expires April 5, 2013 [Page 8]
Internet-Draft pNFS Lustre Layout Operations October 2012
5.1. pnfs_los_targetid_type4
The following enum specifies the manner in which a OST (OSS target) can
be specified. The target can be specified by the network access
protocol type used.
/// enum pnfs_los_targetid_type4 {
/// LOS_TARGET_TCP = 1,
/// LOS_TARGET_IB = 2
/// };
5.2. pnfs_los_deviceaddr4
The specification for an object device address is as follows:
/// union pnfs_los_targetid4 switch(pnfs_los_targetid_type4 oti_type) {
/// case OSS_TARGET_TCP:
/// netaddr4 tcp_addr<>;
///
/// case OSS_TARGET_IB:
/// netaddr4 ib_addr<>;
///
/// default:
/// void;
/// };
///
/// struct pnfs_los_deviceaddr4 {
/// pnfs_los_targetid4 lda_targetid;
/// opaque lda_ossname<>;
/// };
5.2.1. OSS Target Identifier
When "lda_targetid" is specified as an OSS_TARGET_TCP, if the TCIP
network protocol is used or as the OSS_TARGET_IB if Infiniband protocol
is used.
When "lda_targetid" is specified the opaque field MUST be formatted as
the OSS name.
5.2.2. Device Network Address
The network address is given with the netaddr4 type, which
specifies a TCP/IP or IB based endpoint (as specified in NFSv4.1 [3]).
When given, the client SHOULD use it to probe for the OSS device at
the given network address. The client MAY still use other discovery
Faibish et al. Expires April 5, 2013 [Page 9]
Internet-Draft pNFS Lustre Layout Operations October 2012
mechanisms to locate the device using the lda_targetid. In
particular, such an external name service SHOULD be used when the
devices may be attached to the network using multiple connections,
and/or multiple storage fabrics (e.g., TCP or IB).
6. Lustre-Based Layout
The layout4 type is defined in the NFSv4.1 [3] as follows:
enum layouttype4 {
LAYOUT4_NFSV4_1_FILES = 1,
LAYOUT4_OSD2_OBJECTS = 2,
LAYOUT4_BLOCK_VOLUME = 3
LAYOUT4_OSS_OBJECTS = 4,
};
struct layout_content4 {
layouttype4 loc_type;
opaque loc_body<>;
};
struct layout4 {
offset4 lo_offset;
length4 lo_length;
layoutiomode4 lo_iomode;
layout_content4 lo_content;
};
This document defines structure associated with the layouttype4
value, LAYOUT4_OSS_OBJECTS. The NFSv4.1 [3] specifies the loc_body
structure as an XDR type "opaque". The opaque layout is
uninterpreted by the generic pNFS client layers, but obviously must
be interpreted by the Lustre storage layout driver. This section
defines the structure of this opaque value, pnfs_oss_layout4.
6.1. pnfs_lov_mds_md
These are the key file mapping data structures. pnfs_lov_ost_data is
per-stripe data structure. lov_mds_md is per file data structure. The
difference between v1 and v3 is that, v3 supports OST pooling.
/// struct pnfs_lov_ost_data { /* per-stripe data structure */
/// __u64 l_object_id; /* OST object ID */
/// __u64 l_object_seq; /* OST object seq number */
/// __u32 l_ost_gen; /* generation of this l_ost_idx */
Faibish et al. Expires April 5, 2013 [Page 10]
Internet-Draft pNFS Lustre Layout Operations October 2012
/// __u32 l_ost_idx; /* OST index in LOV (lov_tgt_desc->tgts) */
/// };
/// struct pnfs_lov_mds_md_v1 { /* LOV EA mds/wire data */
/// __u32 lmm_pattern; /* LOV_PATTERN_RAID0, LOV_PATTERN_RAID1 */
/// __u64 lmm_object_id; /* LOV object ID */
/// __u64 lmm_object_seq; /* LOV object seq number */
/// __u32 lmm_stripe_size; /* size of stripe in bytes */
/// __u16 lmm_stripe_count; /* num stripes in use for this object
*/
/// __u16 lmm_layout_gen; /* layout generation number */
/// struct pnfs_lov_ost_data lmm_objects[0]; /* per-stripe data */
/// };
///
/// struct pnfs_lov_mds_md_v3 { /* LOV EA mds/wire data */
/// __u32 lmm_pattern; /* LOV_PATTERN_RAID0, LOV_PATTERN_RAID1 */
/// __u64 lmm_object_id; /* LOV object ID */
/// __u64 lmm_object_seq; /* LOV object seq number */
/// __u32 lmm_stripe_size; /* size of stripe in bytes */
/// __u16 lmm_stripe_count; /* num stripes in use for this object
*/
/// __u16 lmm_layout_gen; /* layout generation number */
/// char lmm_pool_name[LOV_MAXPOOLNAME]; /* must be 32bit
aligned */
/// struct pnfs_lov_ost_data lmm_objects[0]; /* per-stripe data
*/
/// };
///
/// union pnfs_lov_mds_md switch (pnfs_lov_magic lmm_magic) {
/// case LOV_MAGIC_V1:
/// pnfs_lov_mds_md_v1 mds_md;
/// case LOV_MAGIC_V3:
/// pnfs_lov_mds_md_v3 mds_md;
/// default:
/// void;
/// };
///
The pnfs_lov_ost_data structure parameterizes the algorithm that
maps a file's contents over the component OST's.
The server MAY grow the file by adding more components to the stripe
while clients hold valid layouts until the file has reached its final
stripe width. The file length in this case MUST be limited to the
number of bytes in a full stripe.
The "pnfs_lov_ost_data" is a per stripe data structure that defines the
location of the stripe in OST and which OST holds the data.
Faibish et al. Expires April 5, 2013 [Page 11]
Internet-Draft pNFS Lustre Layout Operations October 2012
"l_object_id" holds the file data's object ID on the OST.
"l_object_seq" holds the object sequence number which is always 0.
"l_ost_idx" holds the OST's index in LOV, and "l_ost_gen" holds the
OST's index generation.
The "lmm_magic" specifies the format of the returned stripping
information. LOV_MAGIC_V1 is used for pnfs_lov_mds_md_v1, and
LOV_MAGIC_V3 is used for "pnfs_lov_mds_md_v3".
The "lmm_pattern" holds the file's stripping pattern. It can be either
LOV_PATTERN_RAID0 or LOV_PATTERN_RAID1. "lmm_object_id" holds the MDS
object ID. "lmm_object_seq" holds the LOV object sequence number.
"lmm_stripe_size" holds the stripe size in bytes. A file is striped
across multiple OSTs in the same stripe size. The "lmm_stripe_count"
holds the number of OSTs over which the file is striped.
"llm_layout_gen" holds the generation of current layout information.
Clients need to obtain layout generation before IO and check layout
generation after IO. If layout generation is changed, client need to
redo the operations.
The "lmm_objects" is an array of "lmm_stripe_count" members containing
per OST file information. Each element is in form of struct
pnfs_lov_ost_data.
6.2. pnfs_los_layout4
The following is the opaque data in generic layout.
/// struct pnfs_los_layout4 {
/// pnfs_lov_magic lmm_magic;
/// pnfs_lov_mds_md lov_mds_md;
/// uint32_t llo_comps_index;
/// pnfs_los_object_cred4 llo_components<>;
/// };
///
pnfs_lov_magic and lov_mds_md are defined as above [section 6.1].
The "llo_components" is an array of "pnfs_los_object_cred4", containing
credentials that client need to use to connect to OSS's. The
"llo_components" may present all credentials that the client needs to
access each object of the file, in which case "llo_comps_index" is set
to zero. Otherwise if a file has multiple layout segments and different
stripping patterns, "llo_comp_index" is set to the index of the object
composing the file, and "llo_components" MUST have exactly one entry.
Note that the layout depends on the file size, which the client
Faibish et al. Expires April 5, 2013 [Page 12]
Internet-Draft pNFS Lustre Layout Operations October 2012
learns, by doing GETATTR commands to the metadata server. The client
uses the file
size to decide if it should fill holes with zeros or return a short
read. Striping patterns can cause cases where component objects are
shorter than other components because a hole happens to correspond to
the last part of the component object.
6.3. Data Mapping Schemes
This section describes the different data mapping schemes in detail.
The Lustre layout always uses a "dense" layout as described in
NFSv4.1 [3]. This means that the second stripe unit of the file
starts at offset 0 of the second component, rather than at offset
stripe_unit bytes. After a full stripe has been written, the next
stripe unit is appended to the first component object in the list
without any holes in the component objects. From the MDS point of view,
each file is composed of multiple data objects striped on one or more
OSTs.
6.3.1. Simple Striping
A file object's layout information is defined in the extended attribute
(EA) of the inode. Essentially, EA describes the mapping between file
object id and its corresponding OSTs.
For example, if file A has a stripe count of three, then its EA will
look like:
EA ---> <obj id x, ost p>
<obj id y, ost q>
<obj id z, ost r>
stripe size and stripe width
In the above equation obj_id is the object identifier of a file
fragment on the ost p, "stripe size" is the size of each file segment
on one OST and "stripe width" is the number of OST's used. So if the
"stripe size" is 1MB, and the "stripe width" is 3, then this would mean
that [0,1M), [4M,5M) ... are stored as object x, which is on OST p; [1M,
2M), [5M, 6M) ... are stored as object y, which is on OST q; [2M,3M),
[6M, 7M) ... are stored as object z, which is on OST r.
Before reading the file, client will query the MDS and be informed that
it should talk to <ost p, ost q, ost r> for this operation. This
Faibish et al. Expires April 5, 2013 [Page 13]
Internet-Draft pNFS Lustre Layout Operations October 2012
information is structured in so-called LSM, and client side LOV
(logical object volume) is to interpret this information so client can
send requests to OSTs. Here again, the client communicates with OST
through a client module interface known as OSC. Depending on the
context, OSC can also be used to refer to an OSS client by itself.
The mapping from the logical offset within a file (L) to the
component object C and object-specific offset O is defined by the
following equations:
L = logical offset into the file
W = stripe width
S = stripe_size
C = (L-L%S)%W
O = L/W/S+L%S
In these equations, S is the number of bytes in a full stripe or stripe
size. C is an index into the array of components, so
it selects a particular OST device. C count from zero. O is the offset
within the OST that corresponds to the file offset. Note that this
computation does accommodate the fact that an object includes all the
file segments that are located on same OST.
For example, consider an object striped over three devices, <OST0 OST1
OST2>. The stripe_size is 1024KB. The stripe width W is thus 3.
Offset 0KB:
C = (0-0%1)%3 = 0 (OST0)
O = 0/3/1024 + (0%1024) = 0
Offset 1024KB:
C = (1024-(1024%1024))%3 = 1 (OST1)
O = 1024/3/1024 +(1024%1024) = 0
Offset 9000KB:
C = (9000-(9000%1024))%3 = 2 (OST2)
O = 9000/3/1024 + (9000%1024) = 810
Offset 102400KB:
C = (102400-(102400%1024))%3 = 1 (OST0)
O = 102400/3/1024 + (102400%4096) = 33
6.4. RAID Algorithms
Faibish et al. Expires April 5, 2013 [Page 14]
Internet-Draft pNFS Lustre Layout Operations October 2012
This section defines the different redundancy algorithms. Note: The
term "RAID" (Redundant Array of Independent Disks) is used in this
document to represent an array of component OST's that store data for
an individual file. The objects are stored on independent OST-based
storage devices. File data is encoded and striped across the array of
component OST's using algorithms developed for block-based RAID
systems.
6.4.1. PNFS_OST_RAID_0
PNFS_OST_RAID_0 means there is no parity data, so all bytes in the
component objects are data bytes located by the above equations for C
and O. If a component object is marked as PNFS_OST_MISSING, the pNFS
client MUST either return an I/O error if this component is attempted
to be read or, alternatively, it can retry the READ against the pNFS
server.
6.4.2. PNFS_OST_RAID_1
PNFS_OST_RAID_1 means there is no parity data, but each OST is mirrored
to another OST. In this case the component objects are data bytes are
still located by the above equations for C and O. If a component object
is marked as PNFS_OST_MISSING, the pNFS client MUST retry the mirrored
OST and return an I/O error if the mirror component is missing as well
and attempt to be read or, alternatively, it can retry the READ against
the pNFS server.
7. Lustre-Based Creation Layout Hint
The layouthint4 type is defined in the NFSv4.1 [3] as follows:
struct layouthint4 {
layouttype4 loh_type;
opaque loh_body<>;
};
The layouthint4 structure is used by the client to pass a hint about
the type of layout it would like to be created for a particular file.
If the loh_type layout type is LAYOUT4_OSS_OBJECTS, then the loh_body
opaque value is defined by the pnfs_osd_layouthint4 type.
7.1. pnfs_los_layouthint4
/// union pnfs_lov_stripe_count_hint4 switch (bool lsc_valid) {
/// case TRUE:
/// uint32_t lsc_stripe_count;
Faibish et al. Expires April 5, 2013 [Page 15]
Internet-Draft pNFS Lustre Layout Operations October 2012
/// case FALSE:
/// void;
/// }
///
/// union pnfs_lov_stripe_size_hint4 switch (bool lss_valid) {
/// case TRUE:
/// uint32_t lss_stripe_size;
/// case FALSE:
/// void;
/// }
///
/// union pnfs_lov_stripe_offset_hint4 switch (bool lso_valid) {
/// case TRUE:
/// uint32_t lso_stripe_offset;
/// case FALSE:
/// void;
/// }
///
/// union pnfs_lov_stripe_pattern_hint4 switch (bool lsp_valid) {
/// case TRUE:
/// uint32_t lsp_stripe_pattern;
/// case FALSE:
/// void;
/// }
///
/// union pnfs_lov_pool_hint4 switch (bool lp_valid) {
/// case TRUE:
/// string lp_pool_name<>;
/// case FALSE:
/// void;
/// }
///
/// struct pnfs_los_layouthint4 {
/// pnfs_lov_stripe_count_hint4 lov_stripe_count_hint;
/// pnfs_lov_stripe_size_hint4 lov_stripe_size_hint;
/// pnfs_lov_stripe_offset_hint4 lov_stripe_offset_hint;
/// pnfs_lov_stripe_pattern_hint4 lov_stripe_pattern_hint;
/// pnfs_lov_pool_hint4 lov_pool_hint;
/// }
///
This type conveys hints for the desired data map. All parameters are
optional so the client can give values for only the parameters it cares
about, e.g. it can provide a hint for the desired number of mirrored
components, regardless of the RAID algorithm selected for the file.
The server should make an attempt to honor the hints, but it can ignore
Faibish et al. Expires April 5, 2013 [Page 16]
Internet-Draft pNFS Lustre Layout Operations October 2012
any or all of them at its own discretion and without failing the
respective CREATE operation.
Faibish et al. Expires April 5, 2013 [Page 17]
Internet-Draft pNFS Lustre Layout Operations October 2012
8. References
8.1. Normative References
[1] http://www.scribd.com/doc/59271212/Understanding-Lustre-
File-System-Internals
[2] Eisler, M., "XDR: External Data Representation Standard",
STD 67, RFC 4506, May 2006.
[3] Shepler, S., Ed., Eisler, M., Ed., and D. Noveck, Ed.,
"Network File System (NFS) Version 4 Minor Version 1
External Data Representation Standard (XDR) Description",
RFC 5662,January 2010.
[4] IETF Trust, "Legal Provisions Relating to IETF Documents",
November 2008,http://trustee.ietf.org/docs/IETF-Trust-
License-Policy.pdf.
[5] Kent, S. and K. Seo, "Security Architecture for the
Internet Protocol", RFC 4301, December 2005.
[6] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[7] Linn, J., "Generic Security Service Application Program
Interface Version 2, Update 1", RFC 2743, January 2000.
This document was prepared using 2-Word-v2.0.template.dot.
Faibish et al. Expires April 5, 2013 [Page 18]
Internet-Draft pNFS Lustre Layout Operations October 2012
Authors' Addresses
Sorin Faibish (editor)
EMC Corporation
228 South Street
Hopkinton, MA 01748
US
Phone: +1 (508) 249-5745
Email: sfaibish@emc.com
Peng Tao
EMC Corporation
8F, Block D, SP Tower
Tsinghua Science Park
Zhongguancun Dong Road
Beijing 100084
PRC
Phone: +86 (10) 8215 8293
Email: tao.peng@emc.com
Faibish et al. Expires April 5, 2013 [Page 19]