Performance Report

Overview

The eSignet Performance Report provides a comprehensive analysis of the system’s responsiveness, reliability, and scalability under various operational conditions. It captures key performance metrics such as throughput, resource utilization and memory consumption. This report is designed to help stakeholders understand how eSignet performs in real-world scenarios, identify potential bottlenecks, and guide future optimizations to ensure a seamless and secure digital identity experience.

Read more about eSignet: eSignet Documentation Further reading for eSignet integration with MOSIP: MOSIP eSignet | MOSIP Docs 1.2.0

Scope

This performance report focuses specifically on benchmarking and evaluating eSignet v1.4.1 under controlled test conditions using a mock Identity Authentication (IDA) system. The key areas covered within the scope of this report include:

1. Performance testing is limited to eSignet v1.4.1 and does not extend to earlier or future versions.

2. Only OTP-based authentication flows were included in the performance assessment. Other factors such as biometrics or password-based flows were excluded because the performance was conducted with mock ID system, so the change in the auth factor does not impact the response times of user authentication and userinfo exchange.

3. The system was tested to achieve and maintain a constant throughput of 100 requests per second (RPS), which in turn translates to 15 transactions per second (TPS) based on the 7 endpoints required to complete 1 transaction.

4. System was able to maintain the throughput of 100 RPS both with and without intentional delays introduced in the mock IDA system. The goal was to ensure consistent performance while remaining within the predefined Service Level Agreements (SLAs) for response times.

5. All performance bottlenecks, failures, or irregularities encountered during testing were documented. Where applicable, fixes and optimizations were implemented and validated to ensure compliance with performance targets.

Approach

To perform the test run, Apache - JMeter was used as the load testing tool to simulate 106 users, maintaining a consistent throughput of 100 requests per second (RPS) for a duration of 30 minutes.

1. 100,000 sample user data was prepared for the run, if the total number of user samples exceeds 100,000 then the first 100,000 sample user data will be unique, and the 100,001st sample data will be a repeat of the 1st sample data from the entry list.

2. To mimic a real-world scenario, an additional delay of 1000ms was introduced for the send-OTP, auth, and token endpoints in the mock identity system. Consequently, the SLA was updated to 1.5 seconds.

Tools Used

1. Jmeter to simulate the user load.

2. Jprofiler to profile JVM and the application code.

Summary

1. Focus: OTP-based authentication flows only

2. Scenarios tested:

   1. With 1-second induced delay in mock Identity Authentication (IDA) system

   2. Without induced delay

3. Load conditions:

   1. Sustained 100 requests per second (RPS)

   2. 30-minute duration

4. Test environment configuration:

   1. eSignet: 2 pods (1500m CPU, 2250Mi memory each)

   2. Mock ID system: 2 pods (300m CPU, 2250Mi memory each)

5. Results:

   1. Target throughput achieved

   2. All Service Level Agreements (SLAs) consistently met

   3. Excellent performance and stability under high load and delayed response conditions

Test Results Summary

API Endpoint	Response time (no delay)	Response time (with delay)	SLA
/authorization/send-otp	58 ms	1260 ms	≤1500 ms
/authorization/v3/authenticate	79 ms	1258 ms	≤1500 ms
/oauth/v2/token	76 ms	1267 ms	≤1500 ms

Scenarios Considered

The performance run is carried out with below assumptions and considerations:

1. Performance of eSignet APIs is checked against MOCK ID plugin.

2. We have considered the OTP auth factor for the performance run.

SLA:

We are considering that the integrated ID system will take 1.5 secs for below integration points to return back the response:

1. Send OTP (/authorization/send-otp)

2. KYC Auth (/authorization/v3/authenticate)

3. KYC Exchange (/oauth/v2/token)

Module Name	Scenario to be tested	API Endpoint	Accepted Response Time	Weightage
eSignet Pre-requisite: Client ID is generated using eSignet API(/v1/esignet/client-mgmt/oidc-client). eSignet API requires PMS auth Token	User authentication with OTP	/csrf/token → GET /authorization/v2/oauth-details → POST /authorization/send-otp → POST /authorization/v3/authenticate → POST /authorization/auth-code → POST /oauth/v2/token → POST /oidc/userinfo → GET	<=100ms <=100ms <=1.5s <=1.5s <=100ms <=1.5s <=100ms	100%

Test Environment

Deviation from the default deployment

1. Migration from the softhsm to PK12. As part of the performance evaluation, a key deviation from the default deployment was the migration from SoftHSM to PKCS#12 (PK12) for key storage. This change enhanced compatibility and operational efficiency such that eSignet met its SLA targets under both normal and stress conditions.

Software (Under Test)

Following ‘Images’ were under the scope of ‘Performance Testing’:

Modules Segregation

Image ID	Tag Name	Comments
mosipid/esignet:1.4.1	release-1.4.1
mosipid/mock-identity-system:release-0.11.0	release-0.11.0	The changes are tested with mosipqa image and are available in release-0.11.x
mosipid/mock-relying-party-ui:0.9.2	release-0.9.2
mosipid/mock-relying-party-service:0.9.2	release-0.9.2
mosipid/oidc-ui:1.4.1	release-1.4.1

Test Data

Performance data load has been populated before the run to ensure realistic results.

DB	Table Name	Number Of Records/Sample Data
mock identity	uin	100000
	fullName	100000
	emailId	100000
	phoneNumber	100000

Test Design

• Test Duration: 30 mins

• Test Type: Load

• Ramp Up: 3 mins

User distribution among the scenarios

Scenario Name	Module Name	API Endpoint	HTTP Method	SLA(ms) <=	Weightage/Load Distribution	Throughput (RPS)
User with OTP authentication	eSignet-service	/csrf/token	GET	100	100	100
		/authorization/v2/oauth-details	POST	100
		/authorization/send-otp	POST	1500
		/authorization/v3/authenticate	POST	1500
		/authorization/auth-code	POST	100
		/oauth/v2/token	POST	1500
		/oidc/userinfo	GET	100

Resource level configuration

Container name	100RPS
	Number of pods	Resource configuration
eSignet	2	resources: limits: cpu: 1500m memory: 2250Mi requests: cpu: 1500m memory: 2250Mi name: JDK_JAVA_OPTIONS value: '-Xms2250M -Xmx2250M'
mock-identity-system	2	resources: limits: cpu: '300m' memory: 2250Mi requests: cpu: 300m memory: 2250Mi name: JDK_JAVA_OPTIONS value: '-Xms1500M -Xmx1500M'

Test Result

Performance test execution results

Application Name	eSignet
Test Duration	13/03/2025 (30 mins)
Status	Pass

Test Report

1. Test results for 100 RPS for a 30-minute run, simulating the real-time ID system by adding a fixed 1-second processing time for each endpoint (send-OTP, auth, and token).

Scenario Name	Transaction Name	API Endpoint	HTTP Method	100 RPS
				Date : 09/06/2025 (half an hour duration)
				# Samples	Min	Average	90% Line	Max	Error %
User with OTP authentication	S01 T01 GetCsrf	/csrf/token	GET	25969	7	12	17	358	0.00%
	S01 T02 OAuthdetails	/authorization/v2/oauth-details	POST	25968	6	12	17	391	0.00%
	S01 T03 Send OTP	/authorization/send-otp	POST	25968	1017	1095	1260	1719	0.00%
	S01 T04 Authentication	/authorization/v3/authenticate	POST	25952	1025	1101	1258	1754	0.03%
	S01 T05 Autorization	/authorization/auth-code	POST	25934	7	14	19	1026	0.00%
	S01 T06 Token	/oauth/v2/token	POST	25934	1030	1106	1267	1964	0.00%
	S01 T07 Userinfo	/oidc/userinfo	GET	25918	6	12	17	437	0.00%

Metrics

Table of eSignet endpoint metrics

Time chart of 90th percentile response time for eSignet services

Dependent services metrics

Table of mock identity system endpoint metrics and time-chart of 90th percentile response time

CPU Utilisation

Memory Utilisation

Resource Calculator

Please refer the details resource calculator published here for sizing guidelines.

Performance Analysis

KPI Comparison

The key performance metrics from the overall run focused on the response times of the endpoints: send-OTP, KYC-Auth, and KYC-Exchange from the integrated identity system to achieve 100/sec through put is shown below.

End points	Expected response time (ms)	Actual Response time (ms)
send-OTP	1500	1260
KYC-Auth	1500	1258
KYC-Exchange	1500	1267

Bottlenecks

Based on the scope of the performance run, no bottlenecks or issues were observed.

Recommendation

The performance of the system was evaluated based on the defined scope and approach, achieving a consistent throughput of 100 requests per second (RPS) during a 30-minute run. It is recommended to use a Resource Calculator for estimating requirements and scaling for the desired number of users and transactions.

Annexure:

• References:

  • Little's law

  • Little’s Law in Performance Testing