On 1/18/2017 10:54 AM, Mikhail Fedosin wrote:
Hello!

In this letter I want to tell you the current status of Glare project
and discuss its future development within the entire OpenStack community.

In the beginning I have to say a few words about myself - my name is
Mike and I am the PTL of Glare. Currently I work as a consultant at
Nokia, where we're developing the service as a universal catalog of
binary data. As I understand it right, Nokia has big plans for this
service, Moshe Elisha can tell you more about them.

And here I want to ask the community - how exactly Glare may be useful
in OpenStack? Glare was developed as a repository for all possible data
types, and it has many possible applications. For example, it's a
storage of vm images for Nova. Currently Glance is used for this, but
Glare has much more features and this transition is easy to implement.
Then it's a storage of Tosca templates. We were discussing integration
with Heat and storing templates and environments in Glare, also it may
be interesting for TripleO project. Mistral will store its workflows in
Glare, it has already been decided. I'm not sure if Murano project is
still alive, but they already use Glare 0.1 from Glance repo and it will
be removed soon (in Pike afaik), so they have no other options except to
start using Glare v1. Finally there were rumors about storing torrent
files from Ironic.

Now let me briefly describe Glare features:

 * Versioning of artifacts - each artifact has a version in SemVer
format and you can sort and filter by this field.
 * Multiblob support - there can be several files and folders per one
artifact.
 * The ease of creating new artifact types with oslo_versionedobjects
framework.
 * Fair immutability - no one can change artifact when it's active.
 * Multistore support - each artifact type data may be stored in
different storages: images may go to Swift; heat templates may be stored
directly in sql-database; for Docker Contatiners you can use Ceph, if
you want.
 * Advanced sorting and filtering with various operators.
 * Uploaded data validation and conversion with hooks - for example,
Glare may check if uploaded file was a valid Tosca template and return
Bad Request if it's not.

If you're interested, I recorded several demos in asciinema, that
describe how Glare works and present the most useful features. Another
demo about uploading hooks will be recorded and published this week.

So, please tell me what you think and recommend in what direction we
should develop the project. Thanks in advance!

Best,
Mike

Useful links:
[1] Api documentation in rst format:
https://etherpad.openstack.org/p/glare-api
[2] Basic artifact workflow on devstack: https://asciinema.org/a/97985
[3] Listing of artifacts: https://asciinema.org/a/97986
[4] Creating your own artifact type with oslo_vo:
https://asciinema.org/a/97987
[5] Locations, Tags, Links and Folders in Glare:
https://asciinema.org/a/99771


__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


What use cases does Glare make available to Nova that Nova doesn't already get from Glance? In other words, what problems/missing features are there in Nova that can't be solved by Glance but can by Glare?

--

Thanks,

Matt Riedemann


__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Reply via email to