11-12 April 2018
The GÉANT SIG-CISS (Cloudy Interoperable Software Stacks) is proposed to be the successor of the long running TF-Storage task force of GÉANT with a wider scope extending from data storage to general cloud infrastructure software stacks, platforms and research workflows. The SIG brings together those who are building/operating R&E clouds and are willing to share strategy, design, deployment, performance optimization, application integration, interoperability, security and other related information, knowledge and best practices as well as participate in joint efforts aimed at addressing needs of academic environment related to building, operating, brokering cloud services and infrastructure and ensuring their interoperability.
Notes and Actions
2.1 - Raise the profile and visibility of the GÉANT SIG-CISS/OSO joint group within the OpenStack Community. SIG-CISS/OSO is a grassroots, self-support, loosely connected group of individuals mostly using the same or similar software stack (CEPH+OpenStack+Kubernetes) for their cloud services platforms. Concrete actions:
a) Register as a specific user group "SIG-CISS" at the OpenStack User Committee (use CERN and SWITCH as anchors)
b) Ask for a generic template to collect information about cloud deployments and map them out in a nice visual format
c) Investigate opportunities to develop a standard, IPv6 optimized, Ubuntu image with appropriate licenses to be shared across the community
Assigned to: Kalle Happonen, CSC
|Report at the next meeting (TNC18)|
|2.2 - Drop the idea of joining the OpenStack Passport programme due to administrative difficulties vs. tangible benefits. Preferred direction forward is in Action 1a).|
2.3 - Follow up on Action 1.1 from the 1st SIG-CISS meeting: Create a reusable distributed fabric on top of existing storage infrastructures with agreed common best practices and policies. Concrete actions:
a) Try and establish multilateral agreements (aka. barter) of storing copies of replicated data in partners' clouds as the first step to a multi-region Amazon S3 alternative.
b) Draft a 2-pager "specifications for a proof-of-concept" architecture; using the idea of distributed CEPH/RADOS gateways behind a central common front-end.
Assigned to: Jan Ivar Beddari
|Strawman document is being drafted.|
2.4 - Follow up to Action 1.3 from the 1st SIG-CISS meeting: Reference architecture and implementation sandbox for OCM. Concrete actions:
a) Ownership, IPR, processes and governance around the "OCM Group" - currently hosted by GÉANT and led by ownCloud, CERN and GÉANT - must be better formalized. Draft the idea using elements of the GÉANT Greenhouse Service.
b) Working code is needed. Developments are on separate tracks: Nextcloud, CERNbox-PowerFolder, ownCloud. Convergence by TNC18!
c) Global coordination of the OCM interoperability (i.e. policy), compliance (i.e. protocol) and operational aspects (i.e. central directory service) are needed. Peter to raise this within GÉANT the company.
d) We need to identify use cases and drive real usage. Up2U project and the CERNbox-PowerFolder co-operations are promising.
Assigned to: Peter Szegedi and Guido Aben
|Convergence by TNC18!|
|2.5 - EC liaison:|