UTMStack

Articles13 min read

Best SIEM for AWS in 2026: 7 Options and How They Collect Logs

By UTMStack Team

UTMStack blog: Best SIEM for AWS

Search for “AWS SIEM” and you hit a puzzle. Amazon Web Services (AWS) sells GuardDuty, Security Hub and Security Lake, and none of them is sold as a SIEM (security information and event management) tool. So teams on AWS pick a SIEM from someone else. Most comparison lists then skip the part that matters most on AWS: how the logs get out of your account, and what that costs.

This guide compares seven SIEM options for AWS on those questions: how each one collects CloudTrail, GuardDuty and VPC Flow Logs, where it runs, how it charges, and who it suits. We build UTMStack, one of the seven, so weigh our view accordingly. Every other product is described from its vendor’s own documentation, checked in November 2026.

The short answer: there is no single best SIEM for AWS. The best one matches where the rest of your logs live and how you want to pay. Microsoft Sentinel suits companies that run on Microsoft 365. Splunk Enterprise Security suits teams that already know Splunk. Elastic Security suits teams that want to build on search. Datadog Cloud SIEM suits teams already watching their applications in Datadog. Panther suits engineers who want detections written in Python. Wazuh and UTMStack suit teams that want open source they can run themselves. Before you choose, check two things: whether the SIEM reads logs from Amazon S3 or from CloudWatch Logs, and whether it bills by data volume, because AWS logs such as VPC Flow Logs are large.

Four ways AWS logs reach a SIEM

Which path a SIEM uses decides your setup work and part of your AWS bill

S3 bucketMost commonLogs land in S3 and the SIEM fetches new files, often after a queue message. Used by Sentinel, Splunk, Elastic, Wazuh and Panther.
CloudWatch LogsSimple to startThe SIEM reads log groups through the API. Used by UTMStack, and offered by Splunk, Elastic, Wazuh and Panther.
Data Firehose pushStreamingAWS pushes logs to the SIEM's endpoint. Offered by Splunk and Datadog.
Security LakeOne schemaThe SIEM subscribes to AWS's security data lake. Offered by Splunk, Wazuh, Datadog and Panther.
Datadog's other path is a Lambda function that it provides, which forwards from S3 or CloudWatch Logs. Each vendor's documentation is listed under Sources.

What AWS gives you, and why it is not a SIEM

AWS has strong security services, but each does one job. Knowing what each does tells you what a SIEM must add.

AWS serviceWhat it doesWhat it leaves out
CloudTrailRecords API calls; Event history keeps 90 days of management events per RegionNo correlation or alerts across sources
GuardDutyDetects threats from AWS logs using threat intelligence and machine learningSees only AWS data, not your firewalls, laptops or identity provider
Security HubSince December 2025, correlates signals from GuardDuty, Inspector, Macie and posture checks; the older service is now Security Hub CSPMBuilt for AWS risks; AWS describes it as integrating with your SIEM
Security LakeCollects security data into S3 buckets in your account, in the Open Cybersecurity Schema Framework (OCSF)A data lake, with no detection rules of its own
OpenSearch Service, Security AnalyticsRuns detection rules over logs you load and sends alertsClosest to a SIEM, but you build and run the collection, retention and response around it
CloudTrail LakeSQL queries over stored CloudTrail eventsClosed to new customers since May 31, 2026

What none of them does is the full SIEM job. That job is to take in logs from AWS and from everything outside it, correlate them, keep them for years, and give one place to investigate. AWS’s own Security Hub FAQ describes integrating with “existing security tools, including SIEM”. For which AWS events matter most, see our guide to AWS CloudTrail SIEM use cases.

How we compared them

We looked at five things for each option:

  1. Collection. How AWS logs reach it, using the four paths in the figure above.
  2. Where it runs. Software as a service (SaaS) only, or also on servers you control.
  3. Pricing model. Per gigabyte (GB), per event, per workload, per device, or free. We name prices only where the vendor publishes them.
  4. Detection content. Whether rules for AWS come ready-made.
  5. Fit. The kind of team and estate it suits.

At a glance

SIEMRuns whereHow it collects AWS logsPricing modelBest for
Microsoft SentinelSaaS on AzureS3 plus queue, with an AWS role that trusts SentinelPer GB ingested, pay-as-you-go or daily commitmentMicrosoft 365 and Azure shops
Splunk Enterprise SecuritySaaS or self-hostedAdd-on (S3 plus queue, CloudWatch, Security Lake) or Firehose pushIngest or workload; activity-based in the cloudTeams with Splunk skills
Elastic SecuritySaaS or self-hostedS3 plus queue, S3 polling, or CloudWatch LogsFree self-managed tier; serverless per GB ingested and retainedTeams that want to build on search
Datadog Cloud SIEMSaaSFirehose, Forwarder Lambda, or Security LakePer million events analyzedTeams already on Datadog
PantherSaaS, or in your own AWS accountS3 with notifications, queue, or CloudWatch LogsNot publishedDetections as Python code
WazuhSelf-hosted, or Wazuh CloudS3 buckets, CloudWatch Logs, Security LakeFree open source; cloud by agent countOpen source, endpoint focus
UTMStackSelf-hosted, or dedicated cloudCloudWatch LogsFree open source; paid per deviceSmall and midsize teams, MSPs, regulated firms

1. Microsoft Sentinel

How it collects AWS logs: the Amazon Web Services S3 connector reads logs that AWS writes to an S3 bucket. An Amazon Simple Queue Service (SQS) queue signals new files, and an AWS Identity and Access Management (IAM) role that trusts Sentinel grants the access. Microsoft’s connector list includes CloudTrail, plus separate connectors for load balancer and CloudFront logs.

Pricing: by data volume. The analytics tier is pay-as-you-go per GB, or a daily commitment from 100 GB to 50,000 GB, which Microsoft says saves up to 52%. A cheaper data lake tier holds long-term data. Microsoft is moving Sentinel into the Defender portal, and says the Azure portal will stop supporting it after March 31, 2027.

Suits: companies whose identities and email already live in Microsoft 365 and Entra ID. Some Microsoft data, such as Office 365 audit logs, is free to ingest; AWS logs are not, so budget for VPC Flow Logs.

2. Splunk Enterprise Security

How it collects AWS logs: the Splunk Add-on for AWS pulls CloudTrail, Config, CloudWatch Logs (including VPC Flow Logs), S3 access logs and Security Lake data, with incremental pulls through SQS and S3. You can also push through Amazon Data Firehose to Splunk’s HTTP Event Collector. Splunk Cloud on AWS adds Data Inputs, formerly Data Manager, which builds the AWS setup for you.

Pricing: ingest-based or workload-based for Enterprise Security, plus activity-based pricing for Splunk Cloud. Prices are by quote.

Suits: large teams with Splunk skills. Splunk’s public research site lists more than 2,200 detections across platforms.

3. Elastic Security

How it collects AWS logs: the AWS integration reads S3 through SQS notifications, polls S3 directly, or reads CloudWatch Logs through the API. It covers CloudTrail, GuardDuty, VPC Flow Logs and many other AWS services.

Pricing: self-managed Elastic has a free Basic tier, and the prebuilt detection rules package works at that level. Editing prebuilt rules in place needs an Enterprise subscription; Basic users duplicate them instead. Elastic Security Serverless charges per GB ingested and per GB retained each month, and dropped per-endpoint fees on March 23, 2026.

Suits: teams comfortable running Elasticsearch, or wanting its search on a usage bill. The source code is available under the Elastic License 2.0, SSPL or AGPLv3, and the default download uses the Elastic License.

4. Datadog Cloud SIEM

How it collects AWS logs: through Amazon Data Firehose, which Datadog recommends for high CloudWatch volume, or through its Forwarder Lambda function, which subscribes to S3 buckets or CloudWatch log groups. Datadog can also subscribe to Security Lake.

Pricing: per million events analyzed each month. Datadog’s public price list shows $5 to $7.50 per million, depending on the billing option.

Suits: teams already using Datadog for application and infrastructure monitoring, who want security signals in the same place.

5. Panther

How it collects AWS logs: Panther’s S3 source reads new objects after S3 event notifications reach its queue through Amazon SNS. It also has SQS and CloudWatch Logs sources. Supported AWS logs include CloudTrail, GuardDuty, VPC Flow Logs, Security Hub, Security Lake, WAF and Config.

Pricing: not published. The pricing page describes two ways to run it, Panther Cloud or your own AWS account and data lake, without listing prices.

Suits: security engineers who want rules as Python functions, kept in Git and tested like software.

6. Wazuh

How it collects AWS logs: the AWS module reads S3 buckets for CloudTrail, VPC Flow Logs, Config, GuardDuty, WAF, load balancers and S3 server access logs. It reads CloudWatch Logs, and can subscribe to Security Lake and Security Hub.

Pricing: free to run yourself. Current 4.x releases are licensed under GPL version 2. Wazuh Cloud is priced in tiers by number of agents.

Suits: teams that want open source with strong endpoint agents and the skills to tune it. For how it compares with other open source options, see our Wazuh alternatives guide.

7. UTMStack

How it collects AWS logs: CloudTrail delivers to a CloudWatch log group, and UTMStack reads that group with an IAM user that has the AWS managed CloudWatchLogsReadOnlyAccess policy. The same path brings in Lambda, Fargate, Elastic Beanstalk and RDS logs, and anything else you route to CloudWatch Logs. UTMStack ships 73 AWS detection rules mapped to MITRE ATT&CK, most of them reading CloudTrail events such as disabled logging, root sign-ins and IAM changes.

Pricing: the Community edition is free and open source under AGPL-3.0. Paid editions are priced per device, never by log volume.

Suits: small and midsize teams, managed service providers and regulated firms that want AWS and on-premises logs in one place without a per-GB bill. Details and limits are in the section below.

A quick way to narrow the list

A starting point, not a ranking; every option here can read CloudTrail

Do most of your users, email and devices run on Microsoft 365 and Entra ID?
Yes
Start with Microsoft SentinelMicrosoft data is cheap to bring in; check the cost of VPC Flow Logs
No
Must the logs stay on servers you run, or the bill stay flat as volume grows?
Yes
Self-hosted open sourceWazuh, UTMStack, or self-managed Elastic
No
A SaaS SIEM you already useSplunk, Datadog, Panther or Elastic Serverless
Then compare the shortlist on the collection path and on what one year of your largest AWS log source would cost.

AWS costs that a SIEM adds

The SIEM licence is only part of the bill on AWS. Getting logs out of the account has its own charges.

  • CloudTrail copies. The first copy of management events delivered to S3 is free. Each extra copy costs $2.00 per 100,000 management events, and data events, such as S3 object reads, cost $0.10 per 100,000. A SIEM that needs its own trail can add a charge you did not have.
  • CloudWatch Logs. Logs sent to CloudWatch Logs are charged per GB ingested and per GB stored. VPC Flow Logs and other “vended” logs have tiered prices that start at $0.50 per GB in N. Virginia and fall at volume. Keep CloudWatch retention short when a SIEM keeps the long-term copy.
  • Data transfer. Logs leaving AWS for a SIEM hosted elsewhere count as data transfer out. Since December 2021, the first 100 GB a month across all Regions, except China and GovCloud, is free; above that it is charged per GB.
  • Per-volume SIEM pricing. VPC Flow Logs are often much larger than CloudTrail. On a per-GB or per-event SIEM, filter them or send them only for important networks.

What CloudTrail charges per 100,000 events

US dollars, from the AWS CloudTrail pricing page

Extra copy of management events$2.00
Data events, such as S3 object reads$0.10
First copy of management events to S3Free
Reuse your existing trail for the SIEM where you can. A second trail that copies the same management events is the easiest charge to avoid.

How UTMStack fits, and where it stops

UTMStack takes in AWS logs alongside Windows and Linux agents, firewalls, Microsoft 365 and other sources, and checks each event against more than 600 detection rules as it arrives, before storage. You can run it on an EC2 instance in your own account, on any other Linux server, or as UTMStack Cloud, a dedicated instance per customer on OVHcloud. Retention runs up to five years with searchable cold storage, and the price does not change with log volume. See threat detection and integrations for the details.

The limits matter on AWS:

  • CloudWatch Logs only. UTMStack reads CloudTrail and other AWS logs from CloudWatch Logs. It does not read S3 buckets directly or subscribe to Security Lake, so you pay CloudWatch ingestion on whatever you route there. The setup guide suggests a log group with one day of retention, so you do not also pay to store the data twice.
  • Access keys, not a role. The integration uses an IAM user’s access key with read-only CloudWatch Logs rights. It does not assume an IAM role across accounts, so rotate that key like any other secret.
  • No response inside AWS. UTMStack’s ready-made response flows act on hosts that run its agent: they block an address, disable a local account or isolate a host. None of them deactivates an IAM user or access key, so that step happens in AWS.
  • Data transfer when hosted outside AWS. UTMStack Cloud runs on OVHcloud, so logs leave AWS and count toward data transfer out. Running UTMStack on EC2 in the same Region avoids that.

For a quote, see pricing or ask us. To size the server for a year of logs, use our SIEM sizing calculator. For the wider market beyond AWS, see best SIEM tools.

Frequently asked questions

Does AWS have a SIEM? Not one it sells as a SIEM. GuardDuty detects threats in AWS data, Security Hub prioritizes AWS risks, and Security Lake stores security data in OCSF. Amazon OpenSearch Service’s Security Analytics can run detection rules, but you build the rest. AWS’s own Security Hub material describes connecting to your SIEM, not replacing it.

Is AWS Security Hub a SIEM? No. Security Hub correlates findings from AWS services such as GuardDuty, Inspector and Macie to rank cloud risks. It does not take in logs from your on-premises servers, laptops or other clouds, and it is not built for long-term log search.

Which AWS logs should a SIEM collect first? CloudTrail management events from every Region, then GuardDuty findings, then VPC Flow Logs for important networks, then data events for sensitive S3 buckets. Our AWS CloudTrail SIEM use cases guide lists the detections each source feeds.

Is there a free SIEM for AWS? Yes, if you run it yourself. Wazuh, UTMStack and self-managed Elastic Basic are free to run and can read AWS logs. You still pay for the EC2 instance or server, the storage, and the AWS charges for moving logs. See our free SIEM guide for the trade-offs.

Should the SIEM run inside AWS? It often helps. A SIEM on EC2 in the same Region avoids data transfer out charges and keeps logs in your account. A SaaS SIEM is less work to run but sends the logs out, and usually bills by volume.

Sources