Download presentation
Presentation is loading. Please wait.
1
Software Management and Pilot Services
Better Software for Better Science. Jorge Gomes, LIP WP3 manager and WP4 member INDIGO-DataCloud Final Review Brussels, 16-17/11/2017 RIA
2
WP3 - Software Management and Pilot Services
WP Overview Objectives and tasks Activities Task T3.1 Software quality assurance (lead: IFCA) Task T3.2 Software release and maintenance (lead: INFN) Task T3.3 Pilot services (lead: LIP) Task T3.4 Exploitation (lead: EGI) Summary Metrics and KPIs Final remarks Thanks to: Pablo Orviz Cristina Aiftimiei Mario David Matthew Viljoen Ludek Matyska and all WP3 members Support the software lifecycle, from development to deployment and exploitation. INDIGO-DataCloud Final Review Software Management and Pilot Services
3
Software Management and Pilot Services
WP3 Objectives Tasks Project Objectives T3.1 Software quality assurance Define and monitor the quality assurance process for the developed software i) code style, ii) unit testing, iii) functional and integration testing, iv) documentation, v) code review, vi) automated deployment Collect metrics and Key Performance Indicators (KPIs) Alignment with QA procedures of external Open Source projects to ease upstream contributions Promote and adopt DevOps practices Continuous integration & Continuous delivery T3.2 Software release and maintenance Software Maintenance & Support Coordination of corrective, adaptive, perfective and preventive maintenance activities, defining Software Maintenance, Problem & Change Management processes Provisioning of support for released software Release Management define release plans & schedules, enforce the agreed software release policies to the software services and components Coordinate the implementation of Continuous Integration & Delivery processes Manage the software artefacts & documentation repositories T3.3 Pilot services Provide infrastructures and services needed for the integration and testing of software: Services to support the development, Software Quality Assurance, release management, maintenance and support. Test infrastructure to support software development. Integration infrastructure to support the integration between components and services. Pilot preview infrastructure with the components and services released by INDIGO-DataCloud. T3.4 Exploitation Make the software releases available in the relevant e-infrastructures, fostering adoption & enlarging the user base, Staged rollout of the software in production environments at several providers. Establish and coordinate a Service Providers Board (SPB). Make the software discoverable (e.g. targeting UMD and EGI applications database). Exploitation strategy including a services portfolio, and complement user oriented exploitation. INDIGO-DataCloud Final Review Software Management and Pilot Services
4
Activities and Achievements
INDIGO-DataCloud Final Review Software Management and Pilot Services
5
Task T3.1 Software quality assurance
INDIGO-DataCloud Final Review Software Management and Pilot Services
6
T3.1: SQA Achievements in INDIGO-2 (I)
Unit testing coverage => increasing trend Total average value is 63.34% (70% recommended in D3.1) ~60% of products over the 70% threshold ~80% of products over 50% coverage Low coverage products: several are external contributions (different SQA baseline) INDIGO-DataCloud Final Review Software Management and Pilot Services
7
T3.1: SQA Achievements in INDIGO-2 (II)
Automated deployment Number of products delivering automated deployment solutions has increased Products not fulfilling: mainly clients or libraries => no configuration involved Ansible is mostly used INDIGO-DataCloud Final Review Software Management and Pilot Services
8
T3.1: SQA Achievements in INDIGO-2 (III)
Documentation All the product’s documentation is now available in markdown format, stored in GitHub repositories Treated as code, benefits: Portable: can be trivially moved elsewhere (Code)-reviewed: typos/wrong modifications spotted easily before merging in production Automated rendering: once in production, documentation is automatically rendered as books, online available (GitBook) Code style 100% of the products aligned with community or de-facto standards Functional testing Automated or manual reports provided for each release/update Code review Performed through GitHub’s pull-requests INDIGO-DataCloud Final Review Software Management and Pilot Services
9
T3.1: DevOps adoption (automation) (I)
Products/middleware Delivery is now automated => Continuous Delivery (CD) Step forward from Continuous Integration (CI) scenario in INDIGO-1 Source code is not only validated but 1) automatically packaged and 2) uploaded to the appropriate means of distribution (RPMs/DEBs, Docker images, tarballs)
10
T3.1: DevOps adoption (automation) (II)
User communities User applications’ development and distribution => profit from DevOps practices Expertise gathered from INDIGO core services development and delivery The schema is similar to the Continuous Delivery scenario described: Application’s source code is checked => code style (CI) Docker image is built and published to indigodatacloudapps repository (CD) Application deployment => Ansible role (NEW) Application validation The application is automatically instantiated using the brand new Docker image generated in the previous step Application is executed and the results validated with the expected output Prototype applied to two applications => DisVis and PowerFit Perfect case studies since they build their own applications, not relying on external applications=> access to codebase INDIGO-DataCloud Final Review Software Management and Pilot Services
11
T3.1: DevOps adoption (automation) (III)
DevOps workflow for DisVis app development DevOps pipeline (automated steps): Source code validation Python’s PEP8 Packaging Docker image building using Ansible roles Delivering Upload Docker image to indigodatacloudapps DockerHub repository Application validation Based on fixed inputs and expected results, the application is executed and the results validated Pipeline execution time: <7 minutes INDIGO-DataCloud Final Review Software Management and Pilot Services
12
Task T3.2 Software release and maintenance
INDIGO-DataCloud Final Review Software Management and Pilot Services
13
Software Lifecycle Management
”make the certified software components available as a set of coherent high quality releases, supported by an efficient maintenance process” By: Defining processes.. Planning .. Managing … the results/artefacts INDIGO-DataCloud Final Review Software Management and Pilot Services
14
Lifecycle Management - Journey
D3.3 P2 Submitted Final Results D3.2 P2 Submitted Updated Plan D3.1 P1 Approved Initial Plan
15
INDIGO-DC Timeline End of project (*) - timeline after the end of project - commitments collected from the PTs & respective partners (**) - lightweight INDIGO-DC SOFTWARE STACK PROVIDERS COLLABORATION AGREEMENT being established.
16
INDIGO-1 - Release
17
INDIGO-2 - Release
18
INDIGO-DC - Maintenance (I)
14 Updates providing 174 component releases
19
INDIGO-DC - Maintenance (II)
Corrective Maintenance Priority driven -> GitHub Dashboard Adaptive Maintenance Porting to new platforms: CentOS7, Ubuntu & 16.04 Openstack Liberty & Newton OpenNebula 4.14 & 5.2 Perfective Maintenance Improve software functionality according to users’ needs Preventive Maintenance Code styles checking & Introduction of regression testing INDIGO-DataCloud Final Review Software Management and Pilot Services
20
INDIGO-DC - Maintenance (III)
INDIGO-DataCloud Final Review Software Management and Pilot Services
21
INDIGO-DC - Maintenance (IV)
65 EMT phone-conf over 30 months INDIGO-DataCloud Final Review Software Management and Pilot Services
22
Activities after INDIGO-DC
Activity During INDIGO-DC After INDIGO-DC INDIGO-DC products Support & Maintenance INDIGO-DC PTs PTs are providing continued support and maintenance to user communities via their SU & new EU projects Continuous Integration & Delivery Jenkins PTs internal tools & Standard tools of upstream distributions (like OpenStack) (*) Maintenance coordination EMT New EU projects - in particular XDC & DEEP-H for dev. & bug fixes, EOSC-Hub for maintenance (**) Release coordination New EU projects Support coordination Direct relationships and SLAs between PTs and user communities Support can continue to be provided through GGUS & GitHub Issues Repositories INDIGO-DC repositories Upstream repositories & Docker Hub INDIGO-DC repository (maintained active), EGI CMD and UMD repos (*) - new projects agreed to continue support of CI & CD using & sharing the same tool (Jenkins) & infrastructure (provided by INDIGO). (**) - commitments and future projects supporting development and/or maintenance. INDIGO-DataCloud Final Review Software Management and Pilot Services
23
Software Lifecycle Management
2 Major Releases 14 Updates 65 EMT phone-confs 196 source code repositories Packages: CentOS7, Ubuntu14.04, 16.04 OpenStack Liberty & Newton OpenNebula 4,14 & 5.2 ~ 730 registered issues (98 open, 633 closed) GGUS:- 7 tickets INDIGO-DataCloud Final Review Software Management and Pilot Services
24
Software Management and Pilot Services
Task T3.3 Pilot services INDIGO-DataCloud Final Review Software Management and Pilot Services
25
T3.3: Status of Pilot preview testbed I
Consolidation of resources into the pilot preview testbed: Several endpoints used for development are now part of the pilot preview. Increase the number of resources in the testbed for larger scale testing: IaaS level: 5 Openstack production sites, including T-Systems plus 2 Kubernetes, Marathon clusters. Several CDMI-QoS endpoints with diverse underlying storage technologies: HPSS, dCache, S3, StoRM, NFS The testbed remains operational after the end of project: Enabling development, testing and demonstration. Basis for new projects. INDIGO-DataCloud Final Review Software Management and Pilot Services
26
T3.3: Pilot preview map # VMs (or hosts) 520 # VCPUs 2460
RAM (GB) 6300 Storage (TB) 100 # Public IPs 90 T3.3: Pilot preview map DESY CDMI server dCache Oneprovider T-Systems Openstack -> PaaS Kubernetes Oneprovider INFN-Padova IAM connector Synergy OOI nova-docker KIT CDMI server Oneprovider PSNC CDMI server IFCA/SIC ooi IAM connector Java-syncrepos Oneprovider Cloud InfoProvider CNAF/INFN IAM Oneprovider CDMI server Orchestrator CloudProviderRanker Zabbix-wrapper SLAManager CMDB Partition Director (DynPart) LIP/INCD OOI IAM connector nova-docker ONEDock rOCCI server NOW ONE-FaSS WaTTS Java-syncrepos Cloud-info-provider IM Onezone + Onaprovider FG API server FG Portal LiferayIAM Indigo Kepler Ophidia CMCC Ophidia web server INFN-Bari IAM connector CDMI server Oneprovider Kubernetes: Marathon Chronos Mesos UPV IM CLUES
27
T3.3: Pilot preview https://ophidialab.cmcc.it:11732/
indigo-ui.ncg.ingrid.pt infra01.ncg.ingrid.pt
28
T3.3: Inclusion of public cloud providers
Several public cloud providers have been integrated into the Pilot Preview infrastructure: T-Systems Open Telekom Cloud, Amazon AWS and Microsoft Azure. Used as part of the overall INDIGO-DataCloud platform services, at the IaaS level and Federated Data Storage/Management. Used to perform stress and scalability tests by the use cases. Examples of success: IM (Infrastructure Manager) and Orchestrator: Automated instantiation of stacks of Virtual Machines. Instantiation of Docker orchestration platforms, such as Mesos clusters (through the use of Tosca templates). Onedata: T-Systems Open Telekom Cloud: Integration of a Onedata Oneprovider service instance into Pilot Preview infrastructure. Storage spaces for WP2 users have been granted to support several use cases INDIGO-DataCloud Final Review Software Management and Pilot Services
29
T3.3: Elasticity and stress tests
Galaxy cluster use case Scalability tests performed on deploying Mesos clusters at several IaaS sites, including public cloud providers Automatic stack deployment of Galaxy portals with elastic clusters, or HTCondor clusters. Clusters of ~100 Virtual Machines; Configured, contextualised, all necessary services in less than one hour. ~100 small sized clusters of about 10 VMs and 100 cores Deployed on Openstack IaaS sites. 4 distinct Virtual HTCondor cluster continuously running. Open Telekom Cloud: The largest cluster with > 700 worker nodes clusters: > 180 VMs,> 1400 cores, ~ 2.7TB of RAM. Using maximum capacity available at the provider No bottleneck found DODAS use case CERN/CMS INDIGO-DataCloud Final Review Software Management and Pilot Services
30
T3.x Security Architecture and design Steering and mgmt of releases Security incidents coordination Technical Board Engineering Management Team Security Group Project bodies Orthogonal to the project activities and across all stages SQA - follow code styles - tests: unit, regression, functional - auto config and install SW RELEASES - security updates - signed artifacts SW MAINTENANCE - vulnerability handling INFRASTRUCTURE - development - pilot preview - commercial providers - testing by use cases - use of TCS certificates - bridging to EGI CSIRT EXPLOITATION - stage rollout - external infrastructures - gathering feedback ASSESSMENT - evaluation of tools - assessment of selected products using SWAMP - assessment of live Onedata deployment Project activities Services and Infrastructure resources INDIGO-DataCloud Final Review Software Management and Pilot Services
31
Software Management and Pilot Services
Task T3.4 Exploitation INDIGO-DataCloud Final Review Software Management and Pilot Services
32
T3.4: Staged rollout and Early Adoption
INDIGO Products went through a Staged Rollout and Early Adopter process (See D3.2 and D3.3) Some new components in INDIGO-2 All of the products have received reports from Early Adopters INDIGO products have been deployed on multiple sites Stage Rollout Pilot Preview and development testbed Other external parties INDIGO Products mapped to existing EGI distribution channels CMD, UMD, EGI DockerHub, AppDB Inclusion is ongoing and status is summarized in D3.3 INDIGO-DataCloud Final Review Software Management and Pilot Services
33
T3.4: Staged Rollout workflow
34
T3.4: Exploitation (user communities)
Use case / User community Algæbloom (Lifewatch) 1 Medical Imaging Biobanks (EuroBioImaging) 2 Molecular Dynamics Simulations (INSTRUCT) 3a Astronomical Data Archive (LBT) 3b Archive System for the Cherenkov Telescope Array (CTA) 4 Disvis, Powerfit, Haddock (WeNMR) 5 Climate models inter comparison data analysis (ENES) 6 eCulture Science Gateway/R (eCulture) 8 Galaxy as a Cloud service (ELIXIR-ITA) 9 MOIST-multidisciplinary oceanic information system (EMSO) 10 Data repository platform (DARIAH) 11 Container-based cluster on demand for data analysis (LHC-CMS) Component User community P# Onedata 0, 1, 2, 3a, 3b, 6, 8, 9, 11 IAM 0, 1, 2, 3a, 8, 10, 11 Orchestrator (TOSCA/Mesos&Chronos) 0, 1, 8, 10, 11 FutureGateway 0, 2, 5, 8, 9 Infrastructure Manager 1, 8, 11 udocker, ONEDock, CLUES, docker containers, Ophidia, OMT, Orchent, TTS/WaTTS, TOSCA parser 1, 2, 3a, 4, 5, 8, 9, 11 INDIGO-DataCloud Final Review Software Management and Pilot Services
35
T3.4: Exploitation (e-Infrastructure level)
EGI AppDB VMops Dashboard (using IM and OCCI) proving to be big success in production EGI DataHub (Onedata) developed and maintained as the de facto federating data solution in EGI ICOS, EOPoC and PanCancer (EOSCpilot demonstrator) actively testing Onedata. Others interested. Other products are available for exploitation by user communities in the EGI distribution channels and also planned in EOSC-hub thematic platforms INDIGO-DataCloud Final Review Software Management and Pilot Services
36
T3.4 Facilitating Contributions upstream
Giacinto’s From Presentation T3.4 Facilitating Contributions upstream TOSCA adaptor for JSAGA ( OCCI implementation for OpenStack ( Extended AWS support for rOCCI in OpenNebula. Python and Java libraries for OCCI support. CDMI and QoS extensions for dCache ( Workflow interface extensions for Ophidia ( OpenID Connect Java implementation for dCache ( MitreID ( and OpenID Connect ( libraries FutureGateway ( OpenStack ( Nova Docker Heat OpenID-Connect for Keystone Pre-emptible instances support (under discussion) OpenNebula ( OneDock Patch to support FaSS Infrastructure Manager ( Clues ( Onedata ( Apache Libcloud ( Kepler Workflow Manager ( The processes and procedures implemented by WP3 facilitated contributions and merge in upstream projects Giacinto Donvito INDIGO-DataCloud -- Final Review
37
Software Management and Pilot Services
T3.4: Success stories Components and features pushed upstream Heat-translator, OpenStack AAI and CLIs,... Adoption by industry TOSCA with Heat translator integrated with IBM Bluemix Onedata (T-Systems and ATOS) PaaS services: IM, Orchestrator, etc (T-Systems and ATOS) Community interest due to openness, SQA and GitHub exposure udocker (> 200 stars on GitHub) integrated by 4 products, merged community contributions, used in production by Mastercode collab, DisVis and Powerfit Ophidia used by Lawrence Livermore and Oak Ridge laboratories Products being pushed to e-infrastructure Distribution Channels (EGI) INDIGO-DataCloud Final Review Software Management and Pilot Services
38
Software Management and Pilot Services
Summary INDIGO-DataCloud Final Review Software Management and Pilot Services
39
Software Management and Pilot Services
Metrics and KPIs Metric Value Adherence to a code style standard 100% Adherence to community de-facto standards Percentage of products following well-known, community de-facto standards 96.86% Average number of documented code style guidelines Mean value of code style rules fulfilled by the products 67.66 Average Unit testing coverage 66.69% Average Automation index 100% means having 1) code style, 2) unit testing, 3) functional testing 75% Metric Value Number of products with Documentation in Markup language 100% Number of Automated Deployment modules developed and published 49 Ansible Galaxy roles 10 PuppetForge manifests Number of Continuous Delivery pipelines in Jenkins CI 38 GitHub statistics 196 repositories 84 contributors 12 outside collaborators ~3K pull requests ~1M lines of code(core products) Upstream contributions Commits merged, catalogue addition 6 OpenStack products 1 OpenNebula adoption 1 OpenID Connect certified 9 other DockerHub indigodatacloud 37 repositories indigodatacloudapps 22 repositories INDIGO-DataCloud Final Review Software Management and Pilot Services
40
Software Management and Pilot Services
Processes and services implemented and operated as planned Two major releases as scheduled github Source code Service Providers Board External providers Exploitation Interface with service providers Infrastructures and marketplaces User community feedback Support product exploitation Software Quality assurance Code style checking Unit testing Functional/Integration testing Documentation Code review Process outline Bugtracker Schedule Artifacts SQA Integration infrastructure Gerrithub Gitbook Jenkins Metrics/Viz grimoire SQA results Software Release and Maintenance Release Management Publish release schedules Manage project public repositories Continuous Integration Continuous Delivery Software maintenance & support Change Management process Support to released software. Problem Management process Artifacts repos Support requests Openproject Engineering Management Team Software Release Maintenance Product teams (WP4,5,6) GGUS helpdesk Exploitation outcomes Bugtracker Maintenance and validation Software Maintenance and Support Preview and Development Infrastructures Stage Rollout Adoption by Communities Adoption Upstream Software Quality Assurance Testing by Use cases Deployment by Infrastructures Adoption by Infrastructures Integration in other products INDIGO-DataCloud Final Review Software Management and Pilot Services
41
Better Software for Better Science.
Thank you Better Software for Better Science. @indigodatacloud INDIGO-DataCloud Final Review Software Management and Pilot Services
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.