PNFS server projects

From Linux NFS

(Difference between revisions)
Jump to: navigation, search
Line 3: Line 3:
For each project we'd like to know how much remains to do to meet those requirements.
For each project we'd like to know how much remains to do to meet those requirements.
-
==== files-based projects ====
+
=== files-based projects ===
-
=== spNFS ===
+
==== spNFS ====
Stores backend data in ordinary local disk filesystems (like ext3), with a hybrid user/kernel (fuse-like) design, and IO to the metadata server.  (XXX: link to details, downloads.)
Stores backend data in ordinary local disk filesystems (like ext3), with a hybrid user/kernel (fuse-like) design, and IO to the metadata server.  (XXX: link to details, downloads.)
Line 11: Line 11:
Currently undergoing a redesign.
Currently undergoing a redesign.
-
=== gfs2 ===
+
==== gfs2 ====
Files-based server using gfs2 to share data between metadata server and data servers.
Files-based server using gfs2 to share data between metadata server and data servers.
Line 17: Line 17:
Initial prototype exists.  Has passed some simple tests.  Known to cheat on the protocol somewhat (based on early 4.1 code, and doesn't enforce stateid's on the data server yet).  Crashes have been reported in use.  Nothing known about performance yet.
Initial prototype exists.  Has passed some simple tests.  Known to cheat on the protocol somewhat (based on early 4.1 code, and doesn't enforce stateid's on the data server yet).  Crashes have been reported in use.  Nothing known about performance yet.
-
=== ocfs2 ===
+
==== ocfs2 ====
None exists now.  However, it appears that a simple pNFS implementation is possible using only userland parts of the cluster software (with no or minimal modifications to kernel filesystem code).  So any work done on gfs2 should also apply to ocfs2 with minimal effort (since they share userland infrastructure).
None exists now.  However, it appears that a simple pNFS implementation is possible using only userland parts of the cluster software (with no or minimal modifications to kernel filesystem code).  So any work done on gfs2 should also apply to ocfs2 with minimal effort (since they share userland infrastructure).
-
==== block-based projects ====
+
=== block-based projects ===
Someone at LSI is working on this, but code hasn't been released yet.
Someone at LSI is working on this, but code hasn't been released yet.
-
==== objects-based projects ====
+
=== objects-based projects ===
-
=== exofs ===
+
==== exofs ====
An object-based filesystem intended in part to be used as a pNFS backend.  Exofs is currently merged, and is nfs-exportable.  Intend to support mirroring and raid0 across multiple OSD's.  Unsure of current status.
An object-based filesystem intended in part to be used as a pNFS backend.  Exofs is currently merged, and is nfs-exportable.  Intend to support mirroring and raid0 across multiple OSD's.  Unsure of current status.

Revision as of 19:29, 6 May 2009

This is a list of projects that we know of that could be candidates for inclusion in major linux distributions. That means they will need to run on Linux, be licensed as free/open source software, and have the quality, performance, and usefulness necessary to convince upstream projects (such as the linux kernel) that they're worth the additional code.

For each project we'd like to know how much remains to do to meet those requirements.

Contents

files-based projects

spNFS

Stores backend data in ordinary local disk filesystems (like ext3), with a hybrid user/kernel (fuse-like) design, and IO to the metadata server. (XXX: link to details, downloads.)

Currently undergoing a redesign.

gfs2

Files-based server using gfs2 to share data between metadata server and data servers.

Initial prototype exists. Has passed some simple tests. Known to cheat on the protocol somewhat (based on early 4.1 code, and doesn't enforce stateid's on the data server yet). Crashes have been reported in use. Nothing known about performance yet.

ocfs2

None exists now. However, it appears that a simple pNFS implementation is possible using only userland parts of the cluster software (with no or minimal modifications to kernel filesystem code). So any work done on gfs2 should also apply to ocfs2 with minimal effort (since they share userland infrastructure).

block-based projects

Someone at LSI is working on this, but code hasn't been released yet.

objects-based projects

exofs

An object-based filesystem intended in part to be used as a pNFS backend. Exofs is currently merged, and is nfs-exportable. Intend to support mirroring and raid0 across multiple OSD's. Unsure of current status.

Personal tools