What is a cloud asset inventory?
A cloud asset inventory is a continuously updated record of every resource running across an organisation's cloud accounts, such as compute, storage, databases, networks and identities, with where each one lives, who owns it and how it connects to other resources. It is built by discovery through the cloud providers' APIs, not by manual entry.
Why a cloud asset inventory matters
Every other cloud discipline starts with the same question: what do we run? Security cannot protect a database nobody knows exists. Finance cannot allocate the cost of a cluster with no owner. A recovery plan cannot bring back a dependency it never listed.
In a data centre the answer changed slowly. Servers arrived on a loading dock. In the cloud, a resource is one API call away, and so is its disappearance. Inventories that were accurate on Monday are wrong by Friday unless something keeps them current.
That is why the major control frameworks put inventory first. Asset inventory is the opening control in the CIS Critical Security Controls, and Asset Management is the first category of the Identify function in the NIST Cybersecurity Framework.
What should a cloud asset inventory include?
| Field | Why it matters |
|---|---|
| Provider, account, region | Where the resource lives and who is billed for it |
| Type and service | What it is: a VM, a bucket, a managed database, a role |
| State and last seen | Whether it still exists, and how current the record is |
| Tags and owner | Who is responsible for it |
| Relationships | What it depends on and what depends on it |
| Configuration | The settings security and recovery decisions are made from |
AWS cloud asset inventory
AWS has two native services for this, and they answer different questions.
- AWS Config records the configuration of AWS resources in an account, how they relate to one another and how they were configured in the past. You choose which resource types it records. An aggregator collects Config data from multiple accounts and Regions, or from an AWS Organizations organization, into one account and Region, but only from accounts that have AWS Config enabled.
- AWS Resource Explorer is a search service: you find resources by name, tag or ID. It searches across Regions once you choose an aggregator Region, and across accounts through AWS Organizations. AWS says searching carries no additional charge.
For AWS cloud asset inventory management across many accounts, the usual pattern is Config with an organization-wide aggregator for history and relationships, and Resource Explorer for quick lookups.
Azure cloud asset inventory
Azure Resource Graph lets you query resources at scale across subscriptions, management groups or a whole tenant, using a query language based on Kusto (KQL). It shows the last 14 days of configuration changes. Microsoft describes it as a free service whose queries are throttled per user, and it powers the Azure portal's search and All resources views.
Two points from Microsoft's own documentation are worth knowing: Resource Graph data is not strongly consistent (it is indexed with a short delay), and a query returns only what the caller has at least read access to, without saying the result might be partial.
Google Cloud Asset Inventory
Google Cloud Asset Inventory is Google's metadata inventory. It lists assets in a project, folder or organization, keeps create, update and delete history for up to 35 days, and can export to BigQuery or Cloud Storage. You can subscribe to a feed to be told when assets change.
Google states that it is not designed for real-time querying of assets, and that the relationship content type requires Security Command Center Premium or Enterprise.
Native tools compared
Each provider documents its own tool. This table only restates what those vendor docs say.
| AWS Config | AWS Resource Explorer | Azure Resource Graph | Google Cloud Asset Inventory | |
|---|---|---|---|---|
| What it is | Configuration recorder with relationships and history | Resource search by name, tag, ID | KQL query service over Azure resources | Metadata inventory with search and export |
| Widest scope | AWS Organizations organization, via an aggregator | AWS Organizations, via multi-account search | Tenant | Organization |
| Change history | Past configurations while Config was recording | Not its purpose | Last 14 days | Up to 35 days |
| Freshness | Records on create, change and delete | Eventually consistent | Indexed with a short delay | Eventually consistent; not for real-time queries |
| Resources covered | AWS | AWS | Azure | Google Cloud |
Multi-cloud asset inventory
Every native tool above describes its own provider's resources. If you run on more than one cloud, you have three or four inventories with different schemas, scopes and history windows, and no single place that answers "how many public databases do we have?".
A multi-cloud asset inventory reads each provider's APIs and normalises the results into one model: one identity per resource, the same fields for every cloud, and one set of relationships. The work is mostly in the normalising. A VM, a subnet and a role mean slightly different things on each provider, and an inventory that does not reconcile them counts the same idea three ways.
How to build a cloud asset inventory
- Connect: a read-only credential in each cloud account, subscription or project: an IAM role on AWS, a service principal on Azure, a service account on Google Cloud.
- Enumerate: call the provider's APIs across every region and service to list what exists. Missing a region is the most common blind spot.
- Normalise: give every resource a stable identity, type, account, region, state and tags, so the same resource is not counted twice under different names.
- Relate: record the relationships between resources: what runs inside what, what attaches to what, what can reach what.
- Repeat: re-run on a schedule or on change events, and record when each resource was last seen so deleted resources drop out instead of lingering.
- Check the runs: keep a history of each discovery run. A failed or partial run leaves an inventory that looks complete and is not.
Organization cloud asset inventory: scoping it right
An organization cloud asset inventory has to start at the top of the hierarchy, not at the accounts someone remembered. On AWS that is the AWS Organizations organization, on Azure the tenant or root management group, on Google Cloud the organization node. Starting there means a new account, subscription or project is inside the inventory from the day it is created, instead of the day someone adds it.
Cloud asset inventory vs CMDB
A configuration management database (CMDB) is the system of record for IT services and their configuration items, usually maintained through change processes and reconciliation. A cloud asset inventory is the discovered truth of what the provider says exists right now.
They complement each other. The common failure is treating a CMDB populated by hand or by periodic import as if it were current. In the cloud, the CMDB is most useful when it is fed from discovery, not reconciled against it once a quarter.
Why relationships matter as much as the list
A flat list answers "how many databases do we have?". It does not answer "what breaks if this subnet goes away?", "which roles can read this bucket?" or "which application pays for this volume?". Those questions are about edges between resources, which is why mature inventories store resources as a graph.
Common gaps
- Shadow resources created outside the provisioning pipeline, which never appear in infrastructure-as-code state.
- Orphaned resources: unattached volumes, old snapshots, idle load balancers that outlive the workload they served.
- Unowned resources with no tag or account mapping, which nobody will patch, pay for or recover.
- Stale records that stay in the inventory after the resource is deleted, because nothing checks when it was last seen.
How Onam Security does this
Asset inventory in Onam Security lists every resource discovery has found across AWS, Azure, Google Cloud, Oracle Cloud, Alibaba Cloud, IBM Cloud and Kubernetes, and models the relationships between them as a graph. Discovery is read-only: nothing is installed and nothing is modified. Every security engine, from posture to attack paths, reads from that one inventory.
Next steps
- Asset inventory in Onam Security: the multi-cloud asset list every engine reads from
- What is CSPM?: what security does with the inventory
- What is a cloud attack path?: how the relationships between resources become routes to your data
Frequently asked questions
What is Google Cloud Asset Inventory?
It is Google Cloud's metadata inventory service. It lists assets in a project, folder or organization, keeps create, update and delete history for up to 35 days, exports to BigQuery or Cloud Storage, and can notify you of changes through feeds. It covers Google Cloud resources.
What is the AWS equivalent of Cloud Asset Inventory?
There is no single equivalent. AWS Config records resource configurations, relationships and history, and an aggregator brings several accounts and Regions together. AWS Resource Explorer searches resources by name, tag or ID across Regions and accounts. Many teams use both.
What is the Azure equivalent of Cloud Asset Inventory?
Azure Resource Graph. It queries resources across subscriptions, management groups or a tenant using a Kusto-based query language, and shows the last 14 days of configuration changes.
What is a multi-cloud asset inventory?
One inventory that reads every cloud provider's APIs and puts the results into a single model, so resources from AWS, Azure, Google Cloud and others share the same fields and relationships. The native tools each cover their own provider.
How do you build a cloud asset inventory?
Connect a read-only credential to each account, enumerate resources in every region, normalise them into one schema, record relationships, re-run on a schedule or on change events, and keep a history of each run so a failed run does not pass for a complete inventory.
What is the difference between a cloud asset inventory and a CMDB?
A CMDB is a system of record for IT services and configuration items, typically maintained through change processes. A cloud asset inventory is discovered from provider APIs and reflects what exists right now. The strongest setups feed the CMDB from discovery rather than reconciling the two by hand.
How often should a cloud asset inventory be updated?
Continuously, or as close to it as the provider APIs allow. Cloud resources are created and deleted in minutes, so an inventory refreshed monthly is mostly a historical document. Recording when each resource was last seen is what lets deleted resources drop out.
Does a cloud asset inventory need agents?
No. Cloud resources are described by the provider's control-plane APIs, so a read-only credential is enough to enumerate them. Agents are only needed to see inside a workload (installed packages, running processes), which is a different question from what exists.