Beta

Reports Security

Internet exposure

What anyone on the internet can reach: open ports, load balancers, public buckets, and snapshots and images shared with everyone. A map from the internet to each resource, by port and by kind. These are your infrastructure's settings, not a scan of what runs inside it.

What it shows

The exposures per account and how they moved over time, with how many are critical, how many high and how many resources could not be checked. A network map draws the way in from the internet, through each account, VPC and subnet, to each resource, labelled with what is open on it, with the critical ones marked. Bars by port or way in, a donut by kind, a flow from kind to account to what is open, and a table of the resources with their public addresses, security groups and a ↗ to the console give the detail.

Critical is an admin port (a remote shell or desktop, a container or cluster API, a database, cache or message broker, a file share), all traffic, or data anyone can copy. High is a protocol that sends passwords in plain text (FTP, TFTP, POP3, IMAP, SNMP, LDAP), more than 100 ports opened in one range, or a load balancer's HTTP listener that serves without redirecting to HTTPS. Only what is open to everyone counts: a rule for an office or a partner's range is not an exposure.

Questions it answers

  • What can anyone on the internet reach in our AWS accounts right now?
  • Is any database, SSH or RDP port open to 0.0.0.0/0, and in which VPC?
  • Which S3 buckets are public, and which snapshots or AMIs are shared with every AWS account?
  • Did an exposure appear this week that was not there last week?

What it reads

Once a day, with the inventory and through the read-only role, qloudpulse works out from each account's configuration what anyone on the internet could reach. Nothing is probed or scanned from outside. A resource's port counts only when all of these hold: a security group opens it to the whole internet, its subnet's network ACL lets the traffic in, and the resource has a public address behind an internet gateway. Internet-facing load balancers count with their listeners. Buckets public by policy or ACL once Block Public Access is applied count too, as do EBS snapshots, RDS snapshots and machine images shared with every AWS account. A bucket or snapshot whose own policy refuses the read is listed as not checked, with the reason and the call that was refused, instead of being left out. Not read: CloudFront, API Gateway, Lambda function URLs and IAM.

What to do with it

Close the critical ones today: narrow the security group rule, or move the resource behind a load balancer or a VPN. For an HTTP listener, add the redirect to HTTPS. For a not-checked bucket, its detail says which call was refused and the ↗ opens where the policy is changed. Then watch the trend: a page that stays flat is the one you want. The same read also feeds the infrastructure findings, which name a public database or bucket as a rule, and the alerts.

Related reports

  • AWS infrastructure findings: Unattached volumes, idle Elastic IPs and versions past support, by rule and by account, with the resources that fail.
  • AWS resources: Every resource the inventory found, with its cost, by type and by its Environment and Team tags.
  • AWS Health: AWS's open issues, upcoming scheduled changes and notifications for every account, in one list, with what closed recently.

The hosted plan has every report, and is free during early access: see pricing. Like every report, this one splits and filters by account, environment, service and resource, steps back 13 months, downloads as PDF and CSV and keeps the view in its link: what every report does. All reports.