6.7. Fuel Containerized Control Plane performance report

Abstract

This document includes Fuel CCP control plane performance test results for various environments (from 50 to 350 nodes). All tests have been performed regarding Fuel Containerized Control Plane performance testing

6.7.1. Environment description

For Kubernetes pre-deployment Kargo tool was used. More information about fuel-ccp and how it can be installed can be found in official documentation.

6.7.1.1. Hardware configuration of each server

All servers have 3 types of configuration describing in table below

Description of servers hardware type 1

server

vendor,model

Dell PowerEdge R630

CPU

vendor,model

Intel,E5-2680 v3

processor_count

2

core_count

12

frequency_MHz

2500

RAM

vendor,model

Samsung, M393A2G40DB0-CPB

amount_MB

262144

NETWORK

interface_name s

eno1, eno2

vendor,model

Intel,X710 Dual Port

bandwidth

10G

interface_names

enp3s0f0, enp3s0f1

vendor,model

Intel,X710 Dual Port

bandwidth

10G

STORAGE

dev_name

/dev/sda

vendor,model

raid1 - Dell, PERC H730P Mini
2 disks Intel S3610

SSD/HDD

SSD

size

3,6TB

6.7.1.2. Network configuration of each server

All servers have same network configuration:

Network Scheme of the environment

6.7.2. Test results

6.7.2.1. Test Case 1: Boot and delete server

The following set of results is dedicated to investigate how Nova installed against Kubernetes cluster via fuel-ccp tool is behaving on various scale.

6.7.2.1.1. 200 nodes OpenStack cluster, concurrency 5, 910 iterations

NovaServers.boot_and_delete_server scenario in nova_200_nodes.html

Boot and delete servers Rally scenario (200 nodes)

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

boot_server

4.001

5.001

5.135

7.686

3.785

delete_server

2.346

2.487

2.517

3.769

2.297

6.7.2.1.2. 350 nodes OpenStack cluster, concurrency 5, 1750 iterations

NovaServers.boot_and_delete_server scenario in nova_350_nodes.html

Boot and delete servers Rally scenario (350 nodes)

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

boot_server

12.679

16.853

18.539

71.071

3.899

delete_server

6.261

8.365

8.613

14.747

0.842

6.7.2.2. Test Case 2: Boot and list servers

The following set of results is dedicated to investigate how Nova installed against Kubernetes cluster via fuel-ccp tool is behaving on various scale.

6.7.2.2.1. 150 nodes OpenStack cluster, concurrency 20, 5520 iterations

Tested density: 40 VMs per compute node

NovaServers.boot_and_list_server scenario in nova_150_nodes_20.html

Boot and list servers Rally scenario (150 nodes)

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

list_servers

4.381

7.135

7.69

172.788

0.105

boot_server

16.931

21.05

22.203

102.507

4.717

6.7.2.2.2. 150 nodes OpenStack cluster, concurrency 40, 5520 iterations

Tested density: 40 VMs per compute node

NovaServers.boot_and_list_server scenario in nova_150_nodes_40.html

Boot and list servers Rally scenario (150 nodes)

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

list_servers

4.656

7.607

7.963

166.783

0.188

boot_server

19.773

25.958

29.396

197.92

9.574

6.7.2.2.3. 200 nodes OpenStack cluster, concurrency 5, 910 iterations

Tested density: 5 VMs per compute node

NovaServers.boot_and_list_server scenario in nova_200_nodes.html

Boot and list servers Rally scenario (200 nodes)

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

list_servers

0.272

0.424

0.476

0.644

0.106

boot_server

4.017

5.037

5.13

5.437

3.794

6.7.2.2.4. 200 nodes OpenStack cluster, concurrency 40, 7280 iterations

Tested density: 40 VMs per compute node

NovaServers.boot_and_list_server [2] scenario in nova_200_nodes.html

Boot and list servers Rally scenario (200 nodes)

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

list_servers

4.237

5.944

7.994

130.39

0.111

boot_server

11.978

15.767

17.764

55.431

3.911

6.7.2.3. Test case 3: Keystone authentication

The following set of results is dedicated to investigate how Keystone installed against Kubernetes cluster via fuel-ccp tool is behaving under various requests per second load. For more information the full Rally report can be used: keystone.html. Failed scenarios are related either to the Keystone configuration tuning under Containerized Control Plane repository or to the huge enough RPS being set for all in one Keystone.

6.7.2.3.1. 50 nodes OpenStack cluster, 30 RPS, 12000 iterations

Authenticate.keystone [4] scenario in keystone.html

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

authenticate

0.08

0.102

0.113

1.015

0.06

6.7.2.3.2. 50 nodes OpenStack cluster, 60 RPS, 12000 iterations

Authenticate.keystone [6] scenario in keystone.html

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

authenticate

0.097

0.132

0.147

1.113

0.073

6.7.2.3.3. 50 nodes OpenStack cluster, 90 RPS, 12000 iterations

Authenticate.keystone [5] scenario in keystone.html

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

authenticate

0.128

0.179

0.199

1.111

0.082

6.7.2.3.4. 50 nodes OpenStack cluster, 120 RPS, 12000 iterations

Authenticate.keystone [7] scenario in keystone.html

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

authenticate

0.195

0.269

0.298

1.455

0.096

6.7.2.3.5. 50 nodes OpenStack cluster, 150 RPS, 12000 iterations

Authenticate.keystone scenario in keystone.html

Operation

Median (sec)

90%ile (sec)

95%ile (sec)

Max (sec)

Min (sec)

authenticate

0.478

0.738

0.817

2.024

0.109