Monitor AWS with Sazabi: What Each AWS Log Source Tells You
Last verified on Nov 8, 2026.
AWS spreads its operational record across several services. Who changed what lives in CloudTrail, network decisions in VPC Flow Logs and your own application output in CloudWatch Logs. Each one answers a different question, and the useful incidents usually need two of them side by side.
Sazabi is AI-native observability and security. It ingests all six AWS sources below, so you can search them in one place instead of switching consoles.
What each AWS source tells you
- CloudTrail records API calls: the caller, the source IP address, the service, the action and whether it failed. It is your answer to "who did that".
- CloudWatch Logs holds whatever your services write to log groups. It is the answer to "what did the application say".
- AWS WAF records each request a web ACL inspected, the rule that decided it and the client IP. It answers "who is hitting the edge".
- VPC Flow Logs records accepted and rejected network flows per network interface. It answers "what is talking to what".
- GuardDuty delivers findings with a type and a severity from 1.0 to 10.0. It answers "what has AWS already flagged".
- API Gateway access logs record route, status and latency per request. They answer "which calls fail and for whom".
How the sources fit together
Four of the sources connect through your AWS account. You create one IAM role, and Sazabi sets up a CloudWatch subscription filter on each log group you pick. CloudTrail, WAF and VPC Flow Logs all ride on the same role as the CloudWatch source, so adding the second source is a small step.
GuardDuty works differently. You deploy an EventBridge rule and API destination in each Region, and Sazabi does not assume a role in your account. API Gateway logs reach Sazabi through CloudWatch Logs, either with the native CloudWatch source or an OpenTelemetry Collector. If you already run a collector, see OpenTelemetry.
A worked example: one address across three sources
Start with the network. The flow log source stores every flow log field as a searchable attribute, including aws.vpc.flow.action. Ask for rejected traffic first:
sazabi logs ask "Show VPC flow log records with aws.vpc.flow.action REJECT in the last 30 minutes"Say a single address dominates the results. This is an example record in the default flow log format, with a documentation-range address:
2 123456789010 eni-1235b8ca123456789 203.0.113.50 172.31.16.21 51514 22 6 12 720 1418530010 1418530070 REJECT OKThe fields are version, account ID, interface ID, source address, destination address, source port, destination port, protocol, packets, bytes, start, end, action and log status. Destination port 22 over protocol 6 is SSH over TCP, and a security group or network ACL refused it.
That tells you something is probing SSH. Now ask the same question of the other sources:
- Search WAF logs for
httpRequest.clientIpequal to the address. If it also sent web requests, see which rule terminated them. - Search CloudTrail for
sourceIPAddressequal to the address. A hit with anerrorCodemeans someone tried your AWS API from there, not just your instances. - Check whether GuardDuty produced a finding that names the same resource.
Three misses is a reasonable result: noisy scanning that your security group already handled. A hit in CloudTrail changes the priority. The point is that one identifier ties four sources together.
Cost and volume
Flow logs and WAF logs can be large. AWS bills CloudWatch Logs ingestion for every flow log record and every logged WAF request. For WAF, add a logging filter that keeps blocked and counted requests only. For flow logs, start with the VPCs that face the internet.
Sazabi pricing is separate. See the pricing page for allowances and overage rates.
Where to go next
Pick the source that matches your first question. If you suspect misuse of credentials, start with CloudTrail. If you are watching the edge, start with AWS WAF. The wider group of security sources is on the security category page, and the cloud platform sources are on the cloud platforms page.
Log sources
- Amazon VPC Flow Logs (native log source, cloud platforms): Connect your AWS account so Sazabi creates a flow log for each VPC you pick and streams its network traffic records in through CloudWatch Logs.
- AWS WAF (native log source, security): Connect your AWS account so Sazabi sets up subscription filters on your AWS WAF log groups and streams web ACL logs in real time.
- AWS CloudTrail (native log source, security): Connect your AWS account so Sazabi can subscribe to your trail's CloudWatch Logs group and stream CloudTrail events in real time.
- AWS CloudWatch (native log source, cloud platforms): Connect your AWS account so Sazabi can set up subscription filters on your CloudWatch log groups and stream logs in real time.
- AWS API Gateway (manual log source, cloud platforms): Send AWS API Gateway access and execution logs from CloudWatch Logs to Sazabi to see who calls which API, failed authorizations, and abusive clients.
- AWS GuardDuty (manual log source, security): Send Amazon GuardDuty findings to Sazabi in near real time with an Amazon EventBridge rule and API destination in each Region.
- Amazon VPC Flow Logs (manual log source, cloud platforms): Run an OpenTelemetry Collector with the awscloudwatch receiver to forward the VPC flow logs you publish to CloudWatch Logs to Sazabi.
- AWS WAF (manual log source, security): Run an OpenTelemetry Collector that reads your aws-waf-logs- CloudWatch log groups and forwards AWS WAF web ACL logs to Sazabi.
- AWS CloudTrail (manual log source, security): Run an OpenTelemetry Collector with the awscloudwatch receiver on your trail's CloudWatch Logs group and export CloudTrail events to Sazabi over OTLP.
- AWS CloudWatch (manual log source, cloud platforms): Run an OpenTelemetry Collector with the awscloudwatch receiver and an otlphttp exporter to forward CloudWatch log group data to Sazabi.
What Sazabi detects from these logs today
These detections are on for every production project, with no monitors or alert rules to write. Sazabi opens an issue when one fires.
- Novel event: A kind of error that has never appeared before.
Frequently Asked Questions
Which AWS services can send logs to Sazabi?
Sazabi has log sources for CloudTrail, CloudWatch Logs, AWS WAF, VPC Flow Logs, GuardDuty and API Gateway. The first four connect through an IAM role in your account. GuardDuty uses an EventBridge rule, and API Gateway logs arrive through CloudWatch Logs.
Do I need a separate setup for each AWS source?
The CloudTrail, WAF and VPC Flow Logs sources reuse the cross-account IAM role of the CloudWatch source, so one role serves all of them. GuardDuty needs an EventBridge rule in each Region.
Does AWS charge for sending these logs to Sazabi?
AWS bills CloudWatch Logs ingestion for the log groups these sources use, such as flow logs and WAF logs, on top of Sazabi usage. See the Sazabi pricing page for the Sazabi side.
Where are the setup steps?
Each AWS source has a setup guide in the Sazabi docs, linked from its page below.
Sources
- The AWS capabilities Sazabi offers: native and manual log sources, an MCP connector and a sandbox CLI (accessed Nov 8, 2026)
- CloudTrail is read through the trail CloudWatch Logs delivery with the same cross-account role as the CloudWatch source (accessed Nov 8, 2026)
- GuardDuty findings arrive through an EventBridge rule and API destination in each Region (accessed Nov 8, 2026)
- Flow log fields become searchable attributes; CloudWatch Logs ingestion is billed by AWS on top of Sazabi usage (accessed Nov 8, 2026)
- Flow log record fields, the ACCEPT and REJECT actions and the default format (accessed Nov 8, 2026)
- CloudTrail record fields such as sourceIPAddress, eventName and errorCode (accessed Nov 8, 2026)
- AWS WAF log fields including clientIp, action and terminatingRuleId (accessed Nov 8, 2026)