[OpenStack Days Korea 2016] Track1

Monasca를 이용한 Cloud 모니터링


변상욱 부장 / HP

All flash CEPH 구성 및 최적화


엄주관 / SK 텔레콤

Red Hat enterprise Linux OpenStack Platform


최원영 이사 / Red Hat

Mellanox CloudX – Acceleration for Cloud Performance and EfficientVirtual Networking


정연구 이사 / Mellanox Technologies

생존형 Hybrid Cloud Platform 개발기


송지형 / SK Planet

카카오는 오픈스택 기반으로 어떻게 5000VM을 운영하고 있을까?


하현 / 카카오

[OpenStack Days Korea 2016] Key Note

개회사


장현정 대표 / OpenStack Korea User Group

Welcome to the Algorithm Economy


Mark Collier, COO / OpenStack Foundation

How open HW and SW drives telco infrastructure innovations


이강원 상무 / SK 텔레콤

아이디어 이코노미에서의 하이브리드 클라우드 전략


박관종 상무 / HPE

Innovating OpenStack Network with SDN solution


박성용 CTO / KulCloud

An SDN Pioneer’s Vision of Networking


Guido Appenzeller, CTSO / VMware

Microsoft is Open!


김명호 박사 / Microsoft

Mitaka Design Series

매번 서밋이 끝나고 난뒤 PTL들이 나와서 발표하는 내용들이 올라오죠. 이번에 나오는 Mitaka에서 추가되는 것들에 대한 이야기입니다.

OpenStack Superuser 사이트에 올라온 글들입니다. 그냥 발표자료에 나온것들만 정리했습니다. 더 자세한 내용은 들어가서 동영상과 함께 작성해놓은 글을 보시면 됩니다.

Swift, Ironic

Swift

  • Burning issues
    • Data at Rest Encryption
    • Continuing the work on erasure coding
    • Request for resource usage data (disk utilisation, etc.)
    • Improving docs and auth for Swift SDK and CLI Tools
  • What’s next
    • Encryption
    • Improving the overall experience for end users and operators in various ways
    • Lowering and smoothing latency
  • What matters in Mitaka
    • Continued focus on scalability and performance
    • Interoperability: adding Keystone policy support for role-based access control

Ironic

  • Burning issues
    • General stability: testing, Nova integration
    • Improving HA for Nova integration
  • What’s next
    • Better testing
    • Nova integration: supporting multiple compute nodes
    • Networking support with Neutron
    • RAID Support
  • What matters in Mitaka
    • Scalability
    • Resiliency
    • Third-party testing for all drivers

Ansible, Oslo, Designate

Ansible

  • Burning issues
    • Upgrading OpenStack
    • Image-based deployment of OpenStack
    • Focusing on providing a toolset for operators
  • What’s next
    • Breaking out repository into composable parts
    • Increased focus on quality of testing
    • Provide examples for use cases that are important to deployers
    • Production-ready upgrade framework for OpenStack
  • What matters in Mitaka
    • Reducing complexity
    • Improve modularity by implementing everything

Oslo

  • Burning issues
    • Both old and new Oslo projects were discussed at Summit
    • Focused on operator feedback and ongoing feedback through bug reports
    • Discussed how to work closely with logging and security working groups
  • What’s next
    • Promoting existing projects built over last few cycles (e.g. TaskFlow)
    • Oslo-privsep: New privilege mechanism based on python functions
    • Oslo-tooz: distributed lock management, leader elections, etc.
  • What matters in Mitaka
    • Hardening the messaging infrastructure
    • New Oslo messaging driver using Pika
    • New driver for ZeroMQ and adding driver for Kafka

Designate

  • Burning issues
    • Gaps remaining from the refactoring that occurred between Kilo and Liberty
    • DNSSEC (DNS Security Extensions)
    • Verification that returned records have not been tampered
    • Integration with Nova and neutron for automatic record creation
  • What’s next
    • Incremental zone transfers (IXFR)
    • Enabling pool scheduler
  • What matters in Mitaka
    • Increasing number of DNS server types, breaking down services into smaller services enhances:
      • Modularity (모듈성)
      • Resiliency (탄력성)
      • Scalability (확장성)

OpenStackClient, Documentation

OpenStackClient

  • Burning issues
    • Help output from tool (feel and heavier)
    • Name spacing of commands
      • Only includes 5 services in repo; others are plugins
      • Some plugins want to use the same resource name
    • Making command set consistent
  • What’s next
    • Adding more sets of commands
      • Neutron networking is missing currently
    • switch to using the OpenStack SDK
  • What matters in Mitaka
    • Performance improvements

Documentation

  • Burning issues
    • Information architecture and data typing
    • Moving towards a task-based approach rather than role-based to make approach rather
    • Impact Big Tent on docs approach
    • Rethinking approach to API documentation
    • A lot of feedback on release notes
  • What’s planned for Mitaka
    • Data typing
    • Role to task centric approach: stating with user guides
    • Re-order index pages
    • Ongoing conversion to RST (Restructured Text)
    • Modifying DocImpact tool

Nova, Kuryr

Nova

  • Burning issues
    • Focused on resource model for Nova and scheduler interactions
    • Held several unconference sessions to drill into specific items raised by community members
    • Plan to reduce the number of configuration options available for Nova
  • What’s next
    • Improved live migrate
    • API docs and functionality with service catalog
    • OS VIF: interface between Nova and Neutron for VIF plugins (in support of Neutron Stadium)
    • Continue evolution of scheduler interface
  • What matters in Mitaka
    • Cells V2 impacts scalability and manageability
    • Live migrate work will improve resiliency
    • OS VIF will help with modularity
    • Nova has work planned in all themes (scalability, resiliency, manageability, modularity, interoperability)

Kuryr

  • Burning issues
    • Focused on identifying pain-points around user deployments with container networking
    • Investigating how the flexibility of Neutron could help with the container networking experience
    • Integration work with Magnum and Kolla
  • What’s next
    • Stabilisation through functional tests, CI, gate testing
    • Nested containers use case with Magnum
    • Deployment and packaging
    • Stretch goal: multi-node environments; Docker Swarm, Mesos, Kubernetes support

Neutron, Cinder, Ceilometer

Neutron

  • Burning issues
    • Network Function Virtualisation
    • Container networking support
    • Advanced services (LBaaS, FWaaS, etc.)
    • Discussions on overall performance, reliability and operability
  • What’s next
    • Better protocol support for L3 (IPv6, BGP, etc.)
    • Continue NFV and container networking work through Neutron Stadium
  • What matters in Mitaka
    • Scalability: continuing the work on distributed virtual routing (DVR)
    • Resiliency
    • Manageability: continuing work towards minimally disruptive upgrades
    • Modularity: continue to build software modules that are loosely coupled (Neutron Stadium)

Ceilometer

  • Burning issues
    • Collaborating with other project (Vitrage and CloudKitty)
    • Requirement for real-time alarms: creating more flexible alarm rules and support for horizontal scaling
    • User Interface (UI) discussions and prototypes
    • Sample configuration files
  • What’s next
    • Aodh: add multiple worker support
    • Ceilometer: add caching and batch support for write performance improvement
    • Gnocchi: metric archive sharing to improve scalability for large data-sets
  • What matters in Mitaka
    • Scalability: alarming and storage
    • Modularity: continue to break down services further to allow for better integration from external projects

Cinder

  • Burning issues
    • how to balance being a general-purpose block storage service and taking advantage of features found in SAN arrays
    • Object inheritance for drivers
    • Expanded functional testing and third-party CI
  • What’s next
    • Replication V2 API
    • API micro-versioning
    • Continued focus on Third-Party CI and functional testing
  • What matters in Mitaka
    • Interoperability: API micro-versioning
    • Resiliency: Investigating Active/Active Cinder