Please use this page to highlight particular accomplishments that your project made during the Iruya release. These will be used to help promote the release.
|Project Name||Description of Accomplishment||Why is this significant?||Contact for follow-up questions|
- Creation of a synthetic workload container.
- Sunset of OpenStack API usage.
- The storperf_workloadagent container may be used under any containerized environment to provide a synthetic workload for StorPerf to drive when characterizing containerized storage performance.
- StorPerf will deprecate the use of OpenStack APIs in the future in order to reduce the dependency on OpenStack and make it more portable to other technologies.
- switch to Ceph in our Functest SUTs
- add Neutron features in our SUTs to improve the test case verification
- test all capabilities possible (e.g. vnc_console)
- verify automatically the CNTT-related containers into additional to the classical ones
- offer new testcases: tempest_horizon, tempest_keystone, tempest_cinder, refstack_platform, refstack_object, octavia, xrally_kubernetes
- support CNTT RC (API testing, API and dataplane benchmarking, VNF onboarding and testing)
- verify ONAP WindRiver OpenLab via Functest CI in a VM ("Inception model")
- allow minimal l2-only testing via Rally
- increase NFVI testing coverage in Functest
- publish new Functest CNTT-specific containers already mandatory in CNTT RC
- verify both Orange and WindRiver ONAP Openlabs
- support multiple deployment models: fully centralized (LFN), distributed or mixed of them
- fully support NAT or ssh tunneling model
- support GitlabCI
- fully manage artifacts publication via S3 (e.g. artificats.opnfv.org)
- dump all log compaign easily via S3 (see Kubernetes Conformance)
- partially integrate a generic dashboard based on Cachet
- add behave driver
- enhance Xtesting for the Network Automation Journey (GitlabCI support, Dashboards, new deployment models, etc.)
- offer the rules to smoothly integrate all test cases needed for CNTT
- implement the CNTT Compliance/Certification cookbook
- Containerization: VSPERF now includes multiple containers - Deploy, Test and Results Containers - in its distribution
- New versions of Upstream Software.
- Configuration Wizard
- With containers, now VSPERF can integrate with 'larger/generic' test frameworks such as XTesting/Dovetail.
- Users can use these containers to use VSPERF in either an Interactive more or Auto mode.
- The containers also help user to deploy and run tests from any where, as long as there is connectivity to DUT-node and Traffic-Generator Node.
- The newer version of the upstream software - DPDK, OVS, VPP and Qemu - makes VSPERF more relevant and useful.
- The configuration helps those new to VSPERF to easily understand, configure and run tests.
- Bump upstream components: Openstack Stein, ODL Neon, Ubuntu Bionic, Python3, DPDK 18.11 etc.
- Minor bug fixing
- Performance tuning of cluster nodes
- Get latest functionality, security updates and bug fixes from upstream projects
- Fix integration issues of various components (e.g. Openstack Dashboard)
- Improved IDF/PDF configuration for the current CI PODs
- Support latest Fenix for Telco infrastructure rolling maintenance and upgrade
- Use Python3, all testing to Fuel
- Keep driving Telco needed rolling maintenance and upgrade. Have it ready with ETSI FEAT03 support, standard implementation and CNTT later on.
NFVbench is a full stack dataplane benchmarking tool currently widely deployed in production around the world. The OPNFV Iruya release matches the NFVBench 4.0.0 version (NFVbench is released independently) and adds the following features:
- Adds support for L3 router insertion in the packet path
- Native support for VxLAN benchmarking with per chain latency stats
- Adds support for High Dynamic Range histograms for latency stats
- Full migration to python3
- L3 router insertion allows to benchmark the overhead of an L3 router in the NFV packet path (e.g. OpenStack Neutron Router)
- VxLAN is an important NFV network overlay use case for large scale deployments that cannot use the simpler VLAN encapsulation
- Dataplane benchmarks typically use low accuracy latency reporting (min/max/avg and coarse grain histogram buckets). HDR histograms provide high accuracy across the range (1usec to 10sec) with lossless histograms.
|Alec Hothan |