cloudfoundry/benchmarkbbs

Name: benchmarkbbs

Owner: Cloud Foundry

Description: Diego BBS Benchmark

Created: 2015-09-24 23:53:52.0

Updated: 2018-05-22 21:25:51.0

Pushed: 2018-05-22 21:25:50.0

Homepage: null

Size: 3826

Language: Go

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

BBS Benchmark

Note: This repo is meant to be used inside a GOPATH that points to a locally cloned diego-release. Alternatively you can use the generate-benchmarks-manifest script from diego-release to generate a benchmark deployment manifest. Then, follow these instructions to deploy the benchmark errand and run it:

 deployment /path/to/diego-benchmarks.yml
 deploy
 run errand benchmark-bbs

This test suite simulates the load of a CF + Diego deployment against a Diego BBS API server.

Running the Tests

The following instructions demonstrate how to run the BBS benchmarks against a CF + Diego deployment on BOSH-Lite.

Enable Locket

The BBS Benchmark test suite maintains cell registrations with the Locket service, so Locket must be deployed before running it. This can be done via the -Q option on the Diego manifest-generation script.

Stop the Brain and Bridge VMs

Before running these tests, the Diego Brain, CC-Bridge, and Cell VMs should be stopped to prevent them from interfering with the test data. On a BOSH-Lite deployment, this can be done by running the following BOSH commands:

 stop brain_z1 0
 stop cc_bridge_z1 0
 stop cell_z1 0

Create Configuration JSON

In order to run the benchmark test suite, you will need to create the necessary configuration file.

Example BOSH-Lite configuration file for a MySQL Backend:

> config.json <<EOF

esired_lrps": 5000,
um_trials": 10,
um_reps": 5,
um_populate_workers": 10,
bs_address": "https://10.244.16.2:8889",
bs_client_http_timeout": "10s",
bs_client_cert": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.crt",
bs_client_key": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.key",
ocket_address": "10.244.16.2:8891",
ocket_client_cert_file": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.crt",
ocket_client_key_file": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.key",
kip_cert_verify": true,
ncryption_keys": {
"key1": "a secure passphrase"

ctive_key_label": "key1",
og_level": "info",
og_filename": "test-output.log",
atabase_driver": "mysql",
atabase_connection_string": "diego:diego@tcp(10.244.7.2:3306)/diego"


rt CONFIG=$PWD/config.json

Example BOSH-Lite configuration file for a Postgres Backend:

> config.json <<EOF

esired_lrps": 5000,
um_trials": 10,
um_reps": 5,
um_populate_workers": 10,
bs_address": "https://10.244.16.2:8889",
bs_client_http_timeout": "10s",
bs_client_cert": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.crt",
bs_client_key": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.key",
ocket_address": "10.244.16.2:8891",
ocket_client_cert_file": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.crt",
ocket_client_key_file": "$GOPATH/manifest-generation/bosh-lite-stubs/bbs-certs/client.key",
kip_cert_verify": true,
ncryption_keys": {
"key1": "a secure passphrase"

ctive_key_label": "key1",
og_level": "info",
og_filename": "test-output.log",
atabase_driver": "postgres",
atabase_connection_string": "postgres://diego:admin@10.244.0.30:5524/diego"


rt CONFIG=$PWD/config.json

Run the Tests

Run ginkgo with the following flags:

go -- -config=$CONFIG

Error Tolerance

To change the fractional error tolerance allowed, add the following property to the configuration JSON:

rror_tolerance": 0.025,

Percent Writes

To change the write load on the database, add the following property to the configuration JSON:

ercent_writes": 5.0,

This property specifies the percentage of the total LRPs desired that the benchmarks will attempt to write on each trial.

Local Route Emitters

To simulate the behavior of having local route emitters on each cell, the following property can be specified in the JSON configuration:

ocal_route_emitters": true,

Metrics

To emit metrics to Datadog, add the following properties to the configuration JSON:

atadog_api_key": "$DATADOG_API_KEY",
atadog_app_key": "$DATADOG_APP_KEY",
etric_prefix": "$METRIC_PREFIX",

To save the benchmark metrics to an S3 bucket, add the following properties:

s_access_key_id": "$AWS_ACCESS_KEY_ID",
s_secret_access_key": "$AWS_SECRET_ACCESS_KEY",
s_bucket_name": "$AWS_BUCKET_NAME",
s_region": "$AWS_REGION",

Collected metrics

• FetchActualLRPsAndSchedulingInfos: The time to fetch information about all ActualLRPs and DesiredLRPs known to the BBS.
• NsyncBulkerFetching: The time to fetch information about new DesiredLRPs from the nsync-bulker process.
• RepBulkFetching: The time to fetch a cell's expected ActualLRPs from the BBS.
• RepBulkLoop The time to calculate ActualLRP statistics and enqueue operations based on the results.
• RepClaimActualLRP: The time required to claim an ActualLRP within the BBS.
• RepStartActualLRP: The time required to register an ActualLRP with the BBS as “started”.

Example:

imestamp" : 1466806960,
easurement" : {
"Name" : "Rep's internal gathering of Actual LRPs",
"Info" : {
  "MetricName" : "RepBulkLoop"
},
"Results" : [
  0.048770786
]
"Average" : 0.048770786,
"Smallest" : 0.048770786,
"Largest" : 0.048770786,
"AverageLabel" : "Average Time",
"SmallestLabel" : "Fastest Time",
"LargestLabel" : "Slowest Time",
"Order" : 5,
"Units" : "s",
"StdDeviation" : 0,

Measurement fields:

• Name: The metric name.
• Info: Additional reporter info for this metric.
• Results: The metric results.
• Average, Smallest, Largest: The average, smallest, and largest values in Results.
• AverageLabel, SmallestLabel, LargestLabel: Labels for the average, smallest, and largest values.
• Order: The index of this metric out of all metrics in this run.
• Units: The units of measurement for this metric.
• StdDeviation: The standard deviation of the results.