Rambunct governs a UK professional services company’s Amazon Web Services (AWS) organisation through AWS Control Tower and defines applications as code. Alarms go to email and a ticket, a trail records management events in all Regions, data disks are snapshotted daily and the CRM is checked hourly.
The challenge
A UK professional services company, which is not named here, runs an Amazon Web Services (AWS) organisation of several accounts. Its management account runs its business applications: a CRM, a document hub and telephony. Before the changes, its operations had these gaps:
- About 35% of the management account’s alarms reached nobody, so none could prompt a response
- No trail recorded management events, the record of who changed what
- The data disks of the CRM server and the document hub server had no backup
- Those applications shared the account that governs the organisation
The solution
Rambunct governs the company’s Amazon Web Services (AWS) organisation and deployed the alerting, audit, backup and availability changes from code.
Governance and change through code:
- An AWS Control Tower landing zone in the London Region governs the organisation
- Preventive and detective controls are enabled, all with the status SUCCEEDED
- The business applications are defined as AWS Cloud Development Kit (AWS CDK) code
- A drift policy requires that what the code created is changed only through code, with a diff read before each deploy
Alerting and ticketing:
- One rule sends every alarm that enters the ALARM state, apart from auto scaling’s target tracking alarms, to an encrypted Amazon Simple Notification Service topic
- The topic emails a confirmed subscriber and calls an AWS Lambda function that opens a ticket
Audit, backup and availability:
- A trail records management events in all Regions
- Daily snapshot policies cover the data disks
- An hourly Amazon CloudWatch Synthetics check requests the CRM’s health address
Architecture
Built on Amazon Web Services (AWS) in the London Region, with the load balancer across multiple Availability Zones.
AWS services: AWS Organizations, AWS Control Tower, AWS CloudTrail, Amazon CloudWatch, Amazon CloudWatch Synthetics, Amazon Simple Notification Service, AWS Lambda, Amazon Elastic Compute Cloud (Amazon EC2), Amazon Relational Database Service (Amazon RDS), Amazon Simple Storage Service (Amazon S3), AWS Cloud Development Kit (AWS CDK)
Results
- Routed alarms to an email subscription and a ticket, where about 35% of the management account’s alarms had reached nobody
- Recorded management events in every Region with a trail, where none had been recorded
- Backed up the data disks with daily snapshots, and the first daily snapshots of both disks completed
- Started checking the CRM once an hour
No incident has yet been handled through the alarm route.






