[Gluster-devel] What would you like to see in GlusterFS.next?

Deepak Shetty dpkshetty at gmail.com
Thu Jan 29 08:21:13 UTC 2015 

In recent past, I opened many feature pages (proposals) with openstack
usecase (mainly using Manila as the openstack service) against glusterfs.
Listing them here to see what from these can make it to glusterfs.next. We
definitely need most if not all of these features to gel GlusterFS nicely
(read: semantically) with Manila.

Each of the below feature pages have Usecase section where i document the
openstack usecase gap and hence the need for the said feature in glusterfs.

*Manila requirements:*

1) GlusterFS SSL Cert based access support at subdir granularity (for
glusterfs protocol only)
    -
http://www.gluster.org/community/documentation/index.php/Features/Subdir_support_in_SSL_auth

2) GlusterFS snapshot support at subdir level (for all protocols)
    -
http://www.gluster.org/community/documentation/index.php/Features/snap_support_for_subdir

3) Adding IP based access control in addition to SSL based access control
(for glusterfs protocol only)
    -
http://www.gluster.org/community/documentation/index.php/Features/IP_access_alongside_SSL_auth

4) glusterd: Ability to create/delete volumes on the fly
    -
http://www.gluster.org/community/documentation/index.php/Features/glusterd-intelligent-volume-creation

5) Dynamic volume set support for SSL specific options
    -
http://www.gluster.org/community/documentation/index.php/Features/auto-refresh-volume-set-ssl

6) Turn off 'glusterfs' protocol access only
    -
http://www.gluster.org/community/documentation/index.php/Features/turn-off-glusterfs-proto-access

7) Ability to create a new share from snapshot
    -
http://www.gluster.org/community/documentation/index.php/Features/share-from-snapshot

8) GlusterFS quota with SSL based auth. is broken
    -
http://www.gluster.org/community/documentation/index.php/Features/quota-with-ssl-auth

9) GlusterFS : data shredding feature as part of gluster volume delete
    -
http://www.gluster.org/community/documentation/index.php/Features/data-shred-support

10) Ability to query GlusterFS volume's capabilities
    -
http://www.gluster.org/community/documentation/index.php/Features/glusterfs-query-cap

11) Show usable size in gluster volume info
    -
http://www.gluster.org/community/documentation/index.php/Features/show-usable-size-in-volume-info


*Cinder requirement**:*

1) gluster geo-rep support at file granularity
    -
http://www.gluster.org/community/documentation/index.php/Features/georep-at-file-granularity
    - Just using 'rsync' is primitive and doesn't help long term. We need
to have glusterfs support APi/verbs like failback, failover, sync/swap etc
that are typical in enterprise storage products supporting DR


*Misc (not fully baked ones, hence haven't open feature pages** yet!)**:*

1) GlusterFS proxy server (using containers?)to serve GlusterFS shares over
a private network (for glusterfs protocol only)
     - Disclaimer: Needs more thought and investigation
     - This is similar to Netapp's virtual storage server which is created
to serve each share on a pvt network
       to achive network level isolation between openstack VMs and storage
array/controller. This makes them to
       claim multi-tenancy at network level too.

2) Enabling non-SSL based access (from a diff. IP) with SSL based access
(from a diff IP) at the same time (for glusterfs protocol only)
     - Openstack Manila Usecase: non-SSL based access can be used for m-shr
service mount and SSL. This is
       needed for m-shr housekeeping (update_share_stats being the primary
work) in Manila. SSL based access is
       needed for tenants to access GlusterFS shares.

     - Needs more discussion, as this might be security threat!


thanx,
deepak


On Thu, Jan 29, 2015 at 12:37 PM, Prashanth Pai <ppai at redhat.com> wrote:

> An object interface to GlusterFS would greatly benefit Swift integration
> efforts. Currently swiftonfile project uses FUSE mount to perform I/O on
> GlusterFS volumes. Moving to libgfapi would increase performance if there
> were an object interface. May be something like this:
>
> glfs_put(object_path, data, metadata)
> data, metadata = glfs_get(object_path)
>
> This idea is vaguely presented in two existing feature pages
>
> http://www.gluster.org/community/documentation/index.php/Features/composite-operations#CREATE-AND-WRITE
>
> http://www.gluster.org/community/documentation/index.php/Features/Feature_Smallfile_Perf#object-store_API
>
> I remember Rudra Siva (CC'd) from community working on something similar :
> http://www.gluster.org/pipermail/gluster-devel/2014-November/042762.html
>
> Regards,
>  -Prashanth Pai
>
> ----- Original Message -----
> From: "Vijay Bellur" <vbellur at redhat.com>
> To: "gluster-users Discussion List" <Gluster-users at gluster.org>, "Gluster
> Devel" <gluster-devel at gluster.org>
> Sent: Wednesday, January 28, 2015 5:31:36 PM
> Subject: [Gluster-devel] What would you like to see in GlusterFS.next?
>
> Hi All,
>
> Thanks to everyone who responded to the recent community survey [1], we
> have an idea of what you think would be necessary in GlusterFS. I have
> tried to collate the wishlist of features under appropriate categories
> here [2].
>
> As a continuation of this, the approaching feature freeze for 3.7 & more
> momentum being gained for 4.0 planning, it would be good to understand
> what features you would like to see in the next releases of GlusterFS.
> To re-state the obvious, 3.x releases will continue to be evolutionary
> and 4.0 is planned to be a major release that brings about architectural
> improvements for better scale, ease of use and performance. More details
> on 4.0 planning can be found at [3].
>
> Please chip in with your ideas and we will continue to curate the
> community feature wishlist with what we learn here.
>
> Cheers,
> Vijay
>
> [1]
> http://www.gluster.org/pipermail/gluster-users/2014-October/019219.html
>
> [2] http://goo.gl/MLPao6
>
> [3] http://www.gluster.org/community/documentation/index.php/Planning40
>
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20150129/8c315832/attachment.html>

More information about the Gluster-devel mailing list