17 views
15 minutes ago

How Burgan Bank Türkiye cut the time it takes to diagnose a critical outage from an hour to 15 minutes

Kenan Çiftçi, Unit Manager, System Management, Burgan Bank

When a critical service went down at Burgan Bank Türkiye, working out the cause used to take about an hour. The bank’s network, infrastructure, database, and application specialists each had their own monitoring tool, and each reviewed their own data before the group could agree on where the fault lay. While that happened, alerts kept arriving, many of them triggered by the same underlying problem.

Today the bank’s engineers usually find the root cause of a critical outage in around 15 minutes. The difference comes from a monitoring platform the bank built over the past few years with SolarWinds and IT consultancy Consolit. Instead of tracking servers, storage and networks one component at a time, it tracks the business services that customers and staff rely on.

Growth that made problems harder to trace

Burgan Bank Türkiye is part of Burgan Bank Group, which operates across the Middle East and North Africa and is ultimately owned by KIPCO Group. As the Turkish bank expanded into retail and digital banking, its technology estate grew more complicated. Traditional infrastructure was joined by containerised workloads, microservices and a range of customer-facing digital services.

Its monitoring had not kept pace. The bank was running ManageEngine OpManager, HP Operations Manager, SiteScope and Zabbix side by side, each covering part of its virtualisation, applications, storage, servers and networks. None of these tools could trace a performance problem from the application down through the infrastructure, and none could show which business service or which customers an incident was affecting.

“Our first objective was to establish a single pane of glass for monitoring and to monitor the entire infrastructure from one centralized platform. Having unified observability was essential for us,” said Kenan Çiftçi, Unit Manager, System Management, at Burgan Bank.

The bank also had to work within Turkish banking regulations, which do not allow core operational and infrastructure telemetry to be stored or processed in a public cloud. Any new platform had to run entirely on the bank’s own premises.

How the platform was set up

Burgan Bank Türkiye tested SolarWinds Observability Self-Hosted in a two-month proof of concept before choosing it and installing it on-premises with Consolit.

The project’s central decision was to organise monitoring around business services. To do that without a great deal of manual work, the team connected SolarWinds to the bank’s configuration management database and its IT service management systems. Custom properties tag each monitored asset automatically, which places it in the right logical group and business-service container. Engineers don’t have to keep those groupings up to date by hand, and that is what allowed the bank to map more than 120 critical business services to the infrastructure they run on.

The team also linked the platform to the bank’s application performance monitoring system through an Elastic API. Application-level and end-user transaction data now appear alongside infrastructure health metrics, so engineers investigating a slow transaction can trace it to the server, database, or network segment responsible.

Several other SolarWinds tools cover specific parts of the estate. Network Configuration Manager and Server Configuration Monitor record configuration changes, check security compliance and keep an audit history. NetPath shows how traffic moves across the network, and Quality of Experience metrics measure transaction health. Database Performance Analyzer looks inside the bank’s databases, showing wait times, blocking sessions, deadlocks, costly SQL statements and performance trends over time.

What has changed

Because data from SAN, network, server, application, database and end-user monitoring now sits in one place, engineers can start from the affected business service and work down to the cause, rather than pulling specialists together to compare separate tools. Senior leaders can see which service is affected and how widely. Average root cause analysis for critical outages has fallen from 60 minutes to 15, a 75% reduction in mean time to resolution.

The network operations centre deals with far fewer alerts than before. Service dependency mapping, custom alert rules, dependency analysis, and event correlation prevent a single fault from generating a separate alert in every downstream system. Alert noise is down by about 67%, which saves the operations team six to seven hours a day.

“Reducing alert noise has significantly lowered the manual workload of the NOC team, enabling engineers to focus on incidents with real customer impact and accelerate incident response,” said Çiftçi.

Database teams can identify problem queries, wait types, blocking sessions and anomalies more quickly, and see which business service each issue affects. Historical performance data helps them tune systems before customers run into problems.

The bank has also built custom dashboards for its database, network, infrastructure, application, development and business teams. Each team sees the data most relevant to its work, but it all comes from the same platform, so teams aren’t working from different numbers when troubleshooting a service or checking its health.

The system is entirely self-hosted, so monitoring data, configuration histories, and audit records never leave the bank’s environment, as required by Turkish banking rules.

Leave a Reply

Welcome to

By signing or creating an account you agree with our Code of conduct & Privacy policy