Ankit Anand

Maintainer, SigNoz

About Kiwi

Kiwi is an innovative fintech startup revolutionizing credit access through UPI payments in India. By integrating instant credit solutions with UPI, Kiwi provides seamless and reliable financial services to a rapidly growing user base. With over a million downloads on Google Play and the App Store, Kiwi's mission is to offer a frictionless user experience in a highly competitive market.

The Challenge: Ensuring Real-Time Reliability While Adhering to Compliances

Operating in the fintech space, Kiwi's systems are predominantly real-time. Any downtime or delay can result in immediate user churn, as customers might switch to competitors like Google Pay.

"All our systems are real-time. If something is down, we need to fix it quickly, or we'll just lose the customer. They'll quickly switch to Google Pay or some other reliable source."

— Khushhal Reddy, Senior Backend Developer, Kiwi

Additionally, Kiwi partners with multiple banks that impose strict guidelines on monitoring, logging, and data residency to comply with the Reserve Bank of India's (RBI) regulations.

"Banks have compliance requirements that mandate a robust monitoring and logging stack. If something goes wrong, they ask us for proofs or traces of the user journey. We need to store and report user activities, errors encountered, and steps taken to resolve issues."

— Khushhal Reddy, Senior Backend Engineer, Kiwi

Some of the key challenges that Kiwi team faced were:

• Real-Time System Reliability: Immediate detection and resolution of issues to prevent user churn.
• Regulatory Compliance: Adhering to bank guidelines and RBI regulations on monitoring, logging, and data residency.
• Scalability and Cost-Effectiveness: Implementing a solution that scales with growth without incurring prohibitive costs.
• Quick Integration and Customization: Needing a monitoring solution that could be rapidly integrated and customized by a small team.

Implementing SigNoz For End-to-End Monitoring & Observability

Kiwi chose SigNoz as their observability platform due to its ease of integration, open-source nature, compliance with data residency requirements, and cost-effectiveness.

One thing that we consistently hear from developers and founders is how quick and easy it is to start working with SigNoz.

Seamless Integration and Customization

"When we started using SigNoz, our team was very small. We wanted quick integrations so we could start using it immediately. SigNoz was very easy to integrate and customize on our end. We could set up new alerts or dashboards in just five minutes without external assistance."

— Khushhal Reddy, Senior Backend Engineer, Kiwi

By utilizing OpenTelemetry, Kiwi could instrument their applications easily, collecting logs, metrics, and traces without significant overhead.

Open Source and Cost-Effective

The team at Kiwi are big fans of open-source tools. They have plans to contribute back to the open-source version of SigNoz. Open-source tools are more transparent by nature, and the team preferred SigNoz cloud over solutions like Datadog and New Relic.

"We didn't want to go with expensive tools. SigNoz being open source was a big plus for us. We're big fans of open-source tools, and it opens up an opportunity to contribute back to the project if we want something custom."

— Khushhal Reddy

Compliance with RBI Guidelines on Data Residency Requirements for Fintechs in India

In 2018, RBI came out with a guideline on storing payment system data of Indian users. According to the guidelines, all payment data need to be stored in data centers located in India.

Challenges in choosing an APM tool for Fintech Companies in India

Data residency was critical for Kiwi, and Kiwi chose to partner up with SigNoz as we provide data centers in India.

"There's a very strict guideline that all data you pass to any third party has to be in India. We requested SigNoz to ensure our data is hosted in the India region to comply with RBI regulations."

— Khushhal Reddy

SigNoz's ability to host data within India made it compliant with Kiwi's regulatory needs.

Using SigNoz for End-to-End Observability

Real-Time Monitoring and Alerting

Kiwi relies heavily on SigNoz for real-time monitoring and alerting to maintain system uptime and reliability.

• On-Call Management: All on-call schedules are based on SigNoz alerts. Developers are immediately notified of any fatal errors or system issues.
• Immediate Issue Detection: Rapid alerts enable the team to address problems before they impact users.

"Our on-call schedules are based on SigNoz alerting. As soon as there are fatal errors in our system, we get notified immediately. It's very helpful in maintaining our service reliability."

— Khushhal Reddy

Setting up Alerts on critical Application Logs

Ensuring the reliable delivery of One-Time Passwords (OTPs) is crucial for Kiwi's user authentication and transaction security. However, monitoring OTP delivery can be challenging, especially when third-party OTP services report successful API responses even if users do not receive the OTPs.

"Even if the OTP servers are up, users might not be getting the OTPs. We cannot track it because the API response gives us a success. To address this, we place specific log lines in our code. If these log lines occur too frequently within a short time frame—say, multiple times in five or ten minutes—we know there's something wrong with the OTP system."

— Khushhal Reddy, Backend Developer, Kiwi

By leveraging SigNoz's logging and alerting capabilities, Kiwi sets up alerts based on the frequency of specific log lines associated with OTP requests. When users repeatedly request OTPs in a short period, it indicates potential issues with OTP delivery. SigNoz monitors these patterns and triggers alerts, enabling the team to investigate and resolve OTP service disruptions promptly and ensure a smooth user experience.

Monitoring Third-Party APIs and Banking Integrations

Given their extensive integration with banking APIs, monitoring third-party services is crucial for Kiwi.

"We have dashboards set up to track all calls to third-party APIs. We monitor every request and response, including success rates and error codes like 5xx or 4xx. This helps us identify issues with bank APIs in real-time."

— Khushhal Reddy

SigNoz empowers the Kiwi team to monitor third-party APIs with:

• Custom Dashboards: Kiwi created custom dashboards to visualize API call metrics, success rates, and failure reasons.
• Alerting on API Failures: Immediate alerts are set up for third-party API downtimes, allowing quick communication with banking partners.
• Business Impact Analysis: By tracking API performance, the business team can assess the impact on user experience and take proactive measures.

"If a bank's API is down, we get alerted immediately and can inform our business team and the bank to resolve the issue quickly. This level of monitoring is essential for us."

— Khushhal Reddy

Empowering Business Teams with Insights

Interestingly, the business team at Kiwi also leverages SigNoz dashboards to make informed decisions.

"Our business team uses SigNoz to track customer retention and the user journey. They monitor specific log lines corresponding to different steps in the app. If they see a drop in users reaching a particular step, they know there's an issue to address."

— Khushhal Reddy

Some of the use-cases for business teams in SigNoz:

• Funnel Analysis: Tracking log lines for each step in the user journey helps identify drop-offs or bottlenecks.
• Performance Metrics: Monitoring third-party services that affect user experience, like OTP delivery systems.
• Cross-Verification: Comparing data from SigNoz with tools like Mixpanel to ensure accuracy.

"Sometimes, we notice discrepancies in Mixpanel events. SigNoz provides a backend verification, ensuring that the data we have is accurate and reliable."

— Khushhal Reddy

Reducing API Response Time from 20 Seconds to Milliseconds

Kiwi encountered a critical issue where a simple GET API call was taking 20 seconds instead of milliseconds, impacting user experience.

"We used SigNoz traces to investigate the issue. The database query was responding in milliseconds, but the API was still slow. SigNoz helped us discover that our application threads were getting choked due to asynchronous calls. We fixed the issue at the application level, and the API response time reduced to milliseconds."

— Khushhal Reddy

Steps Taken:

1. Trace Analysis: Used SigNoz's trace explorer to identify the time taken at each span, noticing the discrepancy between DB response time and total API response time.
2. JVM Metrics Monitoring: Checked JVM metrics to identify thread choking issues.
3. Application-Level Fixes: Resolved threading issues in the Java application code.
4. Validation: Monitored the improved performance using SigNoz to confirm the fix.

"Initially, we thought it was a DB problem, but SigNoz helped us pinpoint the real issue in our application code. It saved us a lot of time and effort."

— Khushhal Reddy

Unified Observability Across Environments

Kiwi uses SigNoz in development, staging, and production environments to ensure consistency and reliability.

• Pre-Production Testing: New code is deployed to lower environments and monitored using SigNoz before moving to production.
• Load Testing: SigNoz is used during load and endurance tests to identify system performance under high traffic.
• Continuous Monitoring: Developers monitor their services daily, using SigNoz to ensure everything is functioning as expected.

"We deploy new code to our lower environments, monitor it using SigNoz, and only then move it to production. It helps us ensure everything is working as expected before it goes live."

— Khushhal Reddy

Results: Enhanced Reliability and Performance

Faster Issue Resolution

• Immediate Alerts: Real-time alerts enable the team to act swiftly.
• Efficient Debugging: Unified logs, metrics, and traces in SigNoz streamline the debugging process.
• Reduced Downtime: Quick resolutions prevent user churn and maintain trust.

Improved System Performance

• Optimized APIs: Identified and fixed performance bottlenecks, reducing response times dramatically.
• Resource Utilization: Monitored CPU, memory, and JVM metrics to optimize resource usage.
• Scalability: Prepared the system to handle 10x increases in user traffic confidently.

"We continuously monitor our systems during load tests using SigNoz. It helps us identify choke points and optimize them before scaling."

— Khushhal Reddy

Cost Savings and Better Developer Experience

Kiwi also uses Cloudwatch for logs. But the costs are unpredictable.

"CloudWatch charges based on the queries you run, and developers wouldn't know how much a query would cost until after running it. It was especially concerning during urgent issues when the priority is to fix the problem, not monitor costs. There were instances where our CloudWatch expenses exceeded our EC2 instance costs."

— Khushhal Reddy, Backend Developer, Kiwi

That’s why the team at Kiwi is planning to move everything to SigNoz soon. We believe you shouldn’t be worried about your queries while debugging your application and even if there is a cost associated with making heavy queries it should not come as a surprise.

Looking Ahead: Expanding SigNoz Usage

Kiwi plans to move more of their infrastructure monitoring from CloudWatch to SigNoz, including AWS logs and additional metrics.

"We're starting to send more of our AWS logs to SigNoz. Our goal is to have all our dashboards for every infra component in one place. It not only saves cost but also makes monitoring more efficient."

— Khushhal Reddy

SigNoz has become an indispensable part of Kiwi's observability stack, enabling them to ensure real-time reliability, maintain regulatory compliance, and optimize system performance—all while being cost-effective.

"SigNoz is something we use daily. If I have ten tabs open, six of them are SigNoz. It’s essential for our operations. It helps me sleep well at night knowing that things are working, and if something goes wrong, we're able to quickly resolve it."

— Khushhal Reddy

SigNoz Cloud is the easiest way to run SigNoz. You can sign up here for a free account and get 30 days of unlimited access to all features.