User Tools

Site Tools


projectreqonpharos

This is an old revision of the document!


Brahmaputra Participating Projects Requirements for Pharos (lab infrastructure)

Project Lead What is your purpose for needing lab resources? What configuration/s and capacity do you need (HW/SW)?
for example Apex for example: Tim Rozet, Dan Radez Describe category of use … a) Development and stand-alone testing b) Performance benchmarking c) Deployment testing. Describe some detail .. e.g. we plan to test xyz. We need xyz hardware and xyz software configured in xyz way to run our tests. Are you using labs today (which ones)? Type of resources needed (full metal, virtual)? Type of deployer/s? Total capacity that you anticipate needing? What support do you need (are there special requirements other than a standard Pharos configuration needed)? When do you need resources for rls B (start and end dates).
StorPerf Mark Beierl, Edgar StPierre Development and standalone testing: for creating the StorPerf artifact (installer), Performance benchmarking: for testing the speed of attached storage Requires: access to controller for creating Heat templates and deploying VMs for testing, internet access for downloading VM images, access to Cinder for creating and attaching volumes
bottlenecks Hongbo Tian Bottlenecks benchmarking: for bottlenecks testing for OPNFV Requires: access to controller for creating Heat templates and deploying VMs for testing,access to VMs to deploy ssh to run benchmarks. VM can be allocated in different HW nodes.
OPNFV documentation project Christopher Price Documentation for OPNFV including templates, toolchain and documentation processes. Requires: Simple workload support in CI (build machines) for the documentation toolchain, integration with Gerrit for patch reviewing, and artifact storage of generated docs.
Characterize vSwitch Performance for Telco NFV Use Cases Maryam Tahhan Performance benchmarking: for testing the performance of Vanilla OVS and OVS with DPDK We need the following hardware https://wiki.opnfv.org/wiki/vswitch_work_environment and centOS7 configured in xyz way to run our tests. We also need a physical traffic generator (IXIA) attached to our HW. We are using the Intel hosted Lab today. We need Full metal resources. We don't need any deployers. Total capacity that you anticipate needing (TODO)? When do you need resources for rls B (start and end dates) (TODO).
kvm4nfv Reddy, Raghuveer, Donald, Dugger Development and standalone testing: to test the kvm4nfv project, including the real time latency, the inter-VM communication round trip cost and live migration duration time. Requires: We need 2 HSW+ platform connected with each other. Need exclusively own the two system when the system is running to avoid noise, also it should be a fixed test bed so that we can compare the latency data between patches and releases. We don't need any deployer since openstack is not ready for real time deployment yet. We hope the resource to be available for normal patch testing also, thus no clear start/end dates.
Yardstick Ana Cunha Development of yardstick framework and NFVI test cases, benchmarking for OPNFV infrastructure and testing for KVM (cyclic test), OVS, SFC, SDNVPN, VNFFG, IPv6, HA. Using today:
Ericsson lab, Huawei lab, ZTE lab, China Mobile lab, LF lab.
Requires:
1) Ericsson lab - a) resources: for yardstick: bare metal system, 3 controllers and minimum 2 compute blades, for virtual traffic classifier: 1 controller, 1 compute blade, 1 blade for traffic generator b) Use: Dev (2-3 hours daily) and CI, c) CI: daily (2-3 hours), weekly (sat&sundays, 12+12 hours);
2) Huawei lab - a) resources: for yardstick: bare metal system, minimum 1 controller and minimum 2 compute blades, for virtual traffic classifier: 1 controller, 1 compute blade, 1 blade for traffic generator b) Use: Dev (2-3 hours daily) and CI, c) CI: daily (2-3 hours), weekly (sat&sundays, 12+12 hours);
3) ZTE lab - a) resources: for yardstick: bare metal system, minimum 1 controller and minimum 2 compute blades b) Use: Dev and CI, c) Dev (2 hours daily), CI: daily (2-3 hours), weekly (optional - sat&sundays);
4) China Mobile lab - a) resources for yardstick/HA testing: bare metal system, 3 controllers and minimum 2 compute blades b) Use: Dev ( Yardstick/HA) and CI, c) Dev (2 hours, 2x/week) CI: daily (2-3 hours), weekly (optional - sat&sundays);
5) LF lab - a) resources: for yardstick: bare metal system, minimum 3 controllers and minimum 2 compute blades b) Use: Dev (2 hours, 2x/week) and CI, c) CI: daily (2-3 hours), weekly (sat&sundays, 12+12 hours);
6) For SFC: need lab to run tests, propose Ericsson lab, part of Yardstick-CI daily&weekly;
7) For SDNVPN: need lab to run tests, propose Ericsson lab, part of Yardstick-CI daily&weekly;
8) For KVM: need lab to run tests, resources: for yardstick: bare metal system;
9) For IPv6: need lab to run tests, resources: for yardstick: bare metal system, minimum 1 controller and minimum 2 compute blades;
10) For VNFFG: need lab to run tests, resources: for yardstick: bare metal system, minimum 1 controller and minimum 2 compute blades;
11) For OVS: need lab to run tests, resources: for yardstick: bare metal system, minimum 1 controller and minimum 2 compute blades;
12) For ARM: need lab with ARM-base HW to run Yardstick NFV infrastruture test, resources: for yardstick: bare metal system, minimum 1 controller and minimum 2 compute blades;
General from all PODs: access to internet, pod description (yaml preferred), rights for troubleshooting, cloud installation done, connection to releng dB for test result API, connection to Jenkins server
(Note: the number and deployment of controllers depend on the installer used by the POD)
projectreqonpharos.1445290012.txt.gz · Last modified: 2015/10/19 21:26 by Ana Cunha