100 likes | 104 Views
Scenario 6 enables resource providers to notify parties about VM state changes, allowing users to build automated workflows. Explore message standards, collaborate, and align requirements with available technologies.
E N D
Scenario 6 Peter Solagna – EGI.eu
VM/Resource state change notification What scenario 6 enables: • Resource providers to notify interested parties about changes in the user's VM state or a proposed state change in the VMM. • End user to build automated VM management and reactive workflows in user space. • Standards for the messages format and transport have to be explored
What has been done Kick start of the scenario • Sent an email to the fedclouds ML • Request for collaboration • Request for use case from user communities • Identified the steps for the scenario fulfillment • Created the Wiki Page to collect: • Description of the scenario & steps • Collaborators for the working group • Summary tables for the notifications implemented by the technologies and the standards • Missing: timeline for the scenario fulfillment • Missing: only one collaborator joined Started the collection of notification sys. implementations
Steps of Scenario 6 Information gathering steps • List the notification system adopted by the technology deployed in the testbed • Collect users requirements • List the currently available standards and map them into the technologies implemented in the testbed • Map users requirements into the notification mechanisms available in the testbed • Identify a core set of features. Identify if there are standards eligible to be suggested as common standards. Operative steps • Technology providers should provide the “status change” notifications through standard formats and transport protocols. • User communities should test these notification mechanisms in their workflows, or implementing a client prototype. Missing steps? • …
Standards • I will extend this table by the next teleconf meeting – adding more columns for a comparative view of the features • Please, if you are familiar with other standards extend it adding new rows
Active/Passive notifications • Active: resource provider sends a notification message to the user when the status of an owned VM changes, or when a specific event occurs • Passive: users poll all their VMs for their status • To implement any automated workflow passive notifications should be available through a PI • OpenStack enables active notification • OpenNebula enables only passive notification through the OCCI interface • Others Tech Providers?
What users want? Scenario6, so far, has no support of users requirements • Would the applications workflows implement active/passive VM status monitoring? • About what different users want to be notified? • Status changes • After an amount wall-clock time • Expected downtimes of the service • Would be useful for users to have periodic summary messages, containing statistics about the number of VMs running/suspended/.. • Would it be in scope? • Is it critical for users the encryption of notification messages? • Notifications shouldn’t contain sensitive data • Encrypted payload of the notification messages • HTTPS connection • Other?
After the information gathering • Within the working group (and hopefully with some users representative), analyze the requirements, mapping them on the currently available technologies and standards • Agree on a set of features that are critical and should go into the blueprint • Start to testthose features on the Fedclouds-TF’s testbeds • How to test them? • Dummy “hello world” style? • Any chance to have real use case to test?
Conclusions The work for the scenario is started but it needs to speed up • Collaborators • Now: for the information gathering/analysis process • Near future: for the use cases testing • Collect information about the current notifications system implemented (and planned) • Use cases from users