COF-C03 domain - 20% of the exam

Account Management and Data Governance

Account Management and Data Governance is 20% of the COF-C03 exam. These are the objectives it covers, each with practice questions, with every answer explained.

The domain in numbers

56
Practice questions
5
Objectives
45
Medium
11
Hard

Where people go wrong

  • Handing a task to the neighbouring system role

    The wrong options give a task to the role next door. SYSADMIN is offered for creating users and roles, and SECURITYADMIN for creating databases or managing account parameters. ACCOUNTADMIN is offered for routine work even where the scenario wants top-level privileges avoided. SYSADMIN owns the object-creation branch. SECURITYADMIN administers users, roles and grants account-wide through MANAGE GRANTS. Only ACCOUNTADMIN spans account parameters, replication and billing.

    See it in a sample question
  • Reversing network policy precedence

    The wrong options reverse which list or scope wins. Snowflake evaluates the blocked IP ranges before the allowed ranges, so an address in both is rejected. A policy attached to a user replaces the account-level policy for that user, and the two are not merged into a union. A policy governs the whole account when it is set with ALTER ACCOUNT SET NETWORK_POLICY, not when it is attached to a single user or to PUBLIC.

    See it in a sample question
  • Matching the wrong authentication method to the scenario

    The wrong options swap authentication methods that sound interchangeable. Multi-factor authentication and single sign-on need an interactive step, so an unattended pipeline uses key-pair authentication, with the public key set on the user and the private key held only by the client. A company with an external identity provider needs federated authentication through a SAML2 security integration, and neither key pairs nor a network policy delegate sign-in.

    See it in a sample question
  • Confusing masking policies with row access policies

    The wrong options pick the policy type that acts on the wrong thing. A masking policy transforms a column value at query time and still returns the row. A row access policy decides which rows appear, and its body can join to a mapping table such as ENTITLEMENTS so assignment changes need no policy edit. Row filtering is offered for obfuscating EMAIL and PHONE, and masking is offered for territory filtering; both fail.

    See it in a sample question
  • Choosing the wrong metadata schema for the scope

    ACCOUNT_USAGE, INFORMATION_SCHEMA and ORGANIZATION_USAGE get exchanged in the wrong options. INFORMATION_SCHEMA is live and low latency but covers current objects in one database. ACCOUNT_USAGE carries a latency that can run to a couple of hours, holds up to a year of account-wide history and includes dropped objects. ORGANIZATION_USAGE rolls consumption up across every account in the organisation.

    See it in a sample question

Objectives in this domain

What to study

The official documents this domain's practice questions cite most.

Sample questions from this domain

Free sampleAccount Management and Data Governancemedium

An administrator needs to enable account-level features such as managing organisation account parameters, configuring account-wide replication, and viewing all billing and usage data. Which system-defined role grants the broadest authority required to perform every one of these account-level operations?

  • ASYSADMIN, because owning the object-creation branch gives it the account-wide authority needed to set account parameters and configure replication
  • BSECURITYADMIN, because managing users, roles, and grants extends to managing account parameters, replication settings, and viewing all account billing data
  • CACCOUNTADMIN, because it sits at the top of the role hierarchy and holds the privileges needed to manage account parameters, replication, and billing Correct
  • DORGADMIN, because it manages billing and replication and therefore also configures the account parameters and usage views inside each individual account
Identify ACCOUNTADMIN as the top role that combines security and system administration with account-level control of parameters, replication, and billing. ACCOUNTADMIN is the most privileged system-defined role and inherits both SYSADMIN and SECURITYADMIN; on top of that it holds account-level privileges to manage account parameters, configure replication, and view all billing and usage data, so it is the only role that covers every operation listed.

Why A is wrong: SYSADMIN manages databases, warehouses, and other objects but lacks account-level rights over parameters, replication, and billing, so it cannot cover every listed operation.

Why B is wrong: SECURITYADMIN administers identities and grants but does not control account parameters, replication, or billing, so the claim that its authority extends there is incorrect.

Why C is correct: ACCOUNTADMIN encapsulates SYSADMIN and SECURITYADMIN and adds account-level authority over parameters, replication, and billing, so it is the only role spanning all listed operations.

Why D is wrong: ORGADMIN manages organisation-level tasks like creating accounts, but it does not administer the in-account parameters and usage views described, so it does not span all the listed operations.

Free sampleAccount Management and Data Governancemedium

A security administrator creates a network policy that lists an office subnet in its allowed IP ranges and also lists one specific address from inside that same subnet in its blocked IP ranges. A user connects from that specific blocked address, which falls within the allowed subnet. What happens to the connection attempt?

  • AThe connection is blocked, because Snowflake evaluates the blocked IP ranges before the allowed ranges, so an explicitly blocked address is rejected even when an allowed range contains it Correct
  • BThe connection is allowed, because an address that appears in the allowed ranges is always permitted regardless of any blocked ranges configured on the same network policy
  • CThe connection is allowed, because the more specific single address in the blocked list is overridden by the broader allowed subnet that contains it
  • DThe connection is blocked, because listing any address in the blocked ranges disables every allowed range on that network policy until the block is removed
Recognise that a Snowflake network policy evaluates its blocked IP ranges before its allowed ranges, so an explicitly blocked address is denied even inside an allowed range. Within a single network policy Snowflake applies the blocked IP list before the allowed IP list, so any address present in the blocked ranges is rejected even when it also falls inside an allowed range; this ordering lets an administrator permit a whole subnet while still denying one specific host within it.

Why A is correct: Snowflake checks the blocked list first and the allowed list second, so an address in both is denied; this is exactly how an administrator carves a single host out of an otherwise permitted subnet.

Why B is wrong: This inverts the precedence rule; Snowflake evaluates the blocked list first, so an address sitting in both lists is rejected rather than allowed by the matching allowed range.

Why C is wrong: Snowflake does not apply longest-prefix or specificity ordering between the lists; the blocked range is honoured first, so the broader allowed subnet does not override the explicit block.

Why D is wrong: A blocked entry only denies its own addresses; the remaining allowed ranges still function normally, so the claim that one block disables the whole allowed list is incorrect.

Free sampleAccount Management and Data Governancemedium

An organisation needs a service account that an automated data pipeline will use to connect to Snowflake from a backend server with no human present to approve a push prompt or enter a one-time passcode. The team wants strong authentication that does not depend on an interactive second factor. Which authentication method best fits this unattended service account?

  • AMulti-factor authentication, because enrolling the service account in 'Multi-factor authentication' lets the pipeline complete the second factor automatically on every programmatic connection it makes
  • BKey-pair authentication, because the service account presents a private key that is verified against a registered public key, giving strong non-interactive authentication for an unattended connection Correct
  • CSingle sign-on, because configuring 'Single sign-on' lets the pipeline silently obtain a session from the corporate identity provider without any user typing credentials at all
  • DPassword authentication, because a long randomly generated password stored in the pipeline configuration is functionally the strongest option available for any automated service account
Select key-pair authentication for unattended service accounts because it gives strong authentication without an interactive second factor. Key-pair authentication has the service account sign with an RSA private key that Snowflake verifies against a public key stored on the user object; because no push approval, passcode, or browser redirect is needed, it delivers strong authentication for headless automation where multi-factor authentication and single sign-on would block on an interactive step.

Why A is wrong: 'Multi-factor authentication' requires an interactive second factor such as a push approval or passcode, which an unattended pipeline cannot supply, so it is unsuitable for a headless service account.

Why B is correct: Key-pair authentication uses an RSA private key held by the pipeline and a public key registered on the user, providing strong authentication with no interactive prompt, which exactly suits an unattended service account.

Why C is wrong: 'Single sign-on' is built for interactive browser-based logins redirected to an identity provider, so it does not provide a clean non-interactive path for a headless backend service account.

Why D is wrong: A static password is weaker than a private key and is more exposed in configuration, so it is not the strongest non-interactive option even though it can connect without prompts.

Free sampleAccount Management and Data Governancehard

A regional sales table must show each salesperson only the rows for the territories they are assigned to, and assignments change frequently and are held in a separate ENTITLEMENTS table that maps roles to territory codes. The governance lead wants row filtering to honour this mapping without editing the policy each time an assignment changes. Which approach best satisfies the requirement?

  • ACreate a row access policy whose body joins to the ENTITLEMENTS mapping table to check whether the current role is entitled to the row's territory, then attach that policy to the sales table Correct
  • BCreate a masking policy on the territory column that returns NULL for territories the current role is not assigned to, so unassigned rows are effectively hidden from the salesperson
  • CGrant each salesperson SELECT only on a per-territory secure view and revoke access to the base table, creating and maintaining one secure view for every territory in the organisation
  • DApply a resource monitor scoped to each salesperson's role so that queries returning rows outside their assigned territories are cancelled before the unentitled rows reach the client
Use a row access policy that joins to a mapping table so row visibility tracks an external entitlements table without policy edits. A row access policy body is a boolean expression evaluated per row and may join to a mapping table such as ENTITLEMENTS to decide visibility; because the decision is computed from the mapping at query time, changing an assignment in the mapping table immediately changes which rows a role sees without any change to the policy or to grants.

Why A is correct: A row access policy can query a mapping table in its body, so entitlement changes are picked up automatically from ENTITLEMENTS without altering the policy, which precisely matches the dynamic per-territory filtering required.

Why B is wrong: A masking policy only blanks a column value and still returns the row, so unentitled rows would remain visible with a NULL territory rather than being filtered out as the requirement demands.

Why C is wrong: Per-territory secure views work but require constant view and grant maintenance as assignments change, which is exactly the manual upkeep the governance lead wants to avoid, so it is not the best fit.

Why D is wrong: Resource monitors govern credit usage on warehouses and cannot inspect or filter row contents, so they cannot enforce territory-based row visibility and are the wrong mechanism.

Free sampleAccount Management and Data Governancemedium

A compliance analyst must report on tables that were dropped six months ago and on login activity from the previous quarter, covering every database in the account in a single query. Which schema should the analyst query to obtain this long historical, account-wide record including objects that no longer exist?

  • AINFORMATION_SCHEMA, because it provides account-wide history spanning a full year and retains dropped objects, allowing a single query over tables removed six months ago
  • BACCOUNT_USAGE, because it stores up to a year of account-wide history and retains records of dropped objects, so it can report on tables removed six months ago Correct
  • CTime Travel on each database, because it preserves prior table states for a year and can return the list of tables that were dropped six months ago across the account
  • DThe result cache, because it stores the output of historical queries for a year and so can reconstruct the list of tables that were dropped six months ago account-wide
Choose ACCOUNT_USAGE for long-retention, account-wide history that includes dropped objects and past activity. The ACCOUNT_USAGE schema in the shared SNOWFLAKE database retains around a year of account-wide history and continues to list objects that have been dropped, alongside historical login and query activity; INFORMATION_SCHEMA is limited to current objects in a single database with short retention, so ACCOUNT_USAGE is the correct source for a six-month-old cross-database compliance report.

Why A is wrong: INFORMATION_SCHEMA is scoped to current objects in one database with only short retention and no dropped-object records, so it cannot report on tables removed six months ago.

Why B is correct: ACCOUNT_USAGE keeps roughly a year of account-wide history and includes dropped objects and past login events, so it can satisfy a six-month-old, cross-database compliance report.

Why C is wrong: Time Travel retains prior object states only for a short configurable window, not a year by default, and it operates per object, so it cannot deliver a single account-wide audit of long-dropped tables.

Why D is wrong: The result cache only reuses results of identical recent queries for a short window and holds no catalogue of dropped objects, so it cannot produce a long historical audit report.

Other domains in this exam

See also the COF-C03 cert hub, the study guide, and the cheat sheet.

Examworthy is not affiliated with or endorsed by Snowflake. Original, blueprint-aligned practice material only.