diff options
Diffstat (limited to 'tests/performance')
| -rw-r--r-- | tests/performance/README.md | 121 |
1 files changed, 66 insertions, 55 deletions
diff --git a/tests/performance/README.md b/tests/performance/README.md index 52613d2c..fcb80589 100644 --- a/tests/performance/README.md +++ b/tests/performance/README.md @@ -1,69 +1,80 @@ -# Moon Yardstick and Bottlenecks Performance Tests +# Moon Yardstick/Bottlenecks Performance Tests The main objective of this document is to describe the performance tests for the Moon project/module. -Moon is a security managment platform which provides a set of security functions to project the underlying OPNFV infrastructure and/or VNFs. -Moon is consisted of 2 parts: a master and a set of slaves. The master holds all security-related information and each slave only fetches and holds -related informations for its local usage from master. +Moon is a security management platform which provides a set of security functions to project the underlying OPNFV infrastructure and/or VNFs. +It is consisted of 2 parts: a master and a set of slaves. The master holds all security-related information and each slave only fetches and holds +related information for its local usage from master. -## Moon Master Performance Tests -In this test, we should: +## Master Performance Tests +### Pre-requisite - setup a Moon master service on a physical server -- create a tenant/scope through the Moon master service -- create a MSL security policy with 4 subject security levels and 4 object security levels for this tenant +- create a project in OpenStack/Keystone +- create a MSL PDP with a model of 4 subject security levels and 4 object security levels, the MLS policy will be defined later -- increase N to find the limit of the security policy (implemented in format of a Docker) - - create N users and N resources (VMs in our case) in this tenant - - simulate 2 operation requests per user per second to Moon's authorization endpoint - - gather performance metrics like CPU, memory, network usages - - throught the iteration, determine the capacity limit for one Docker +### Policy Size Test +Increase the number of users and resources N to find the limit of the security policy +- create N users and N resources (VMs in our case) in this MLS security policy +- sends 5 authz requests/second +- gather performance metrics like CPU, memory, network usages +Through the iteration, determine the maximal number of N to support 5 requests/second -- setup 20 user and 20 resources (VMs in our case) for one tenant - - increase the number of tenants to test the maximal number of tenants on the server +### PDP Number Test +- setup 20 user and 20 resources (VMs in our case) for each MLS PDP +- sends 5 authz requests/second for each MLS PDP +- increase the number of PDP to test the maximal number of PDP on the master -- setup 5 tenants of N users and N resources (VMs in our case) in each tenant - - increase N by simulating 2 operation requests per user per second to the Moon's authorization endpoint - - gather performance metrics like CPU, memory, network usages - - throught the iteration, dermine the maximal user/resource number of these 5 tenants/Dockers on the server +### Policy Size Test for 5 PDPs +- setup 5 PDPs of N users and N resources (VMs in our case) +- sends 5 authz requests/second for each MLS PDP +- gather performance metrics like CPU, memory, network usages +Through the iteration, determine the maximal user/resource number of these 5 PDPs -- setup 10 tenants of N users and N resources (VMs in our case) in each tenant - - increase N by simulating 2 operation requests per user per second to the Moon's authorization endpoint - - gather performance metrics like CPU, memory, network usages - - throught the iteration, dermine the maximal user/resource number of these 10 tenants/Dockers on the server +### Policy Size Test for 10 PDPs +- setup 10 PDPs of N users and N resources (VMs in our case) +- sends 5 authz requests/second for each MLS PDP +- gather performance metrics like CPU, memory, network usages +Through the iteration, determine the maximal user/resource number of these 10 PDPs -- setup 20 tenants of N users and N resources (VMs in our case) in each tenant - - increase N by simulating 2 operation requests per user per second to the Moon's authorization endpoint - - gather performance metrics like CPU, memory, network usages - - throught the iteration, dermine the maximal user/resource number of these 20 tenants/Dockers on the server - -## Moon Slave Performace Tests -In this test, we should: -- setup a Moon master service on a physical server -- setup a Moon slave service on a physical server -- create a tenant/scope through the Moon master service -- create a MSL security policy with 4 subject security levels and 4 object security levels for this tenant through the Moon master service +### Policy Size Test for 20 PDPs +- setup 20 PDPs of N users and N resources (VMs in our case) +- sends 5 authz requests/second for each MLS PDP +- gather performance metrics like CPU, memory, network usages +Through the iteration, determine the maximal user/resource number of these 20 PDPs -- increase N to find the limit of the security policy (implemented in format of a Docker) - - create N users and N resources (VMs in our case) in this tenant - - simulate 2 operation requests per user per second to Moon slave's authorizatoin endpoint - - gather performance metrics like CPU, memory, network usages of Moon slave - - throught the iteration, dermine the capacity limit for one Docker of Moon slave - -- setup 20 user and 20 resources (VMs in our case) for one tenant through the Moon slave service - - increate the number of tenants to test the maximal number of tenants on the server of the Moon slave - -- setup 5 tenants of N users and N resources (VMs in our case) in each tenant through the Moon master service - - increate N by simulating 2 operation requests per user per second to the Moon slave's authorization endpoint - - gather performance metrics like CPU, memory, network usages of both Moon master and Moon slave - - throught the iteration, dermine the maximal user/resource number of these 5 tenants/Dockers on the server of Moon slave -- setup 10 tenants of N users and N resources (VMs in our case) in each tenant through the Moon master service - - increate N by simulating 2 operation requests per user per second to the Moon slave's authorization endpoint - - gather performance metrics like CPU, memory, network usages of both Moon master and slave - - throught the iteration, dermine the maximal user/resource number of these 10 tenants/Dockers on the server of the Moon slave +## Master-Slave Performance Tests +### Pre-requisite +- setup a Moon master on a physical server +- setup a Moon slave on a physical server +- create a project in OpenStack/Keystone +- create a MSL PDP with a model of 4 subject security levels and 4 object security levels, the MLS policy will be defined later on the master + +### Slave Policy Size Test +Increase the number of users and resources N to find the limit of the security policy +- create N users and N resources (VMs in our case) in this MLS security policy on the master +- sends 5 authz requests/second to the slave +- gather performance metrics like CPU, memory, network usages of the slave +Through the iteration, determine the maximal number of N to support 5 requests/second of the slave + +### Slave PDP Number Test +- setup 20 user and 20 resources (VMs in our case) for each MLS PDP on the master +- sends 5 authz requests/second for each MLS PDP to the slave +Through the iteration, determine the maximal number of PDP to support 5 requests/second of the slave -- setup 20 tenants of N users and N resources (VMs in our case) in each tenant through the Moon master service - - increate N by simulating 2 operation requests per user per second to the Moon slave's authorization endpoint - - gather performance metrics like CPU, memory, network usages of both Moon master and slave - - throught the iteration, dermine the maximal user/resource number of these 20 tenants/Dockers on the server of the Moon slave +### Slave Policy Size Test for 5 PDPs +- setup 5 PDPs of N users and N resources (VMs in our case) on the master +- sends 5 authz requests/second for each MLS PDP to the slave +- gather performance metrics like CPU, memory, network usages of the slave +Through the iteration, determine the maximal user/resource number of these 5 PDPs +### Slave Policy Size Test for 10 PDPs +- setup 10 PDPs of N users and N resources (VMs in our case) on the master +- sends 5 authz requests/second for each MLS PDP to the slave +- gather performance metrics like CPU, memory, network usages of the slave +Through the iteration, determine the maximal user/resource number of these 10 PDPs +### Slave Policy Size Test for 20 PDPs +- setup 20 PDPs of N users and N resources (VMs in our case) on the master +- sends 5 authz requests/second for each MLS PDP to the slave +- gather performance metrics like CPU, memory, network usages of the slave +Through the iteration, determine the maximal user/resource number of these 20 PDPs |