Cloud IaaS Misconfiguration Review: Lock Down Permissions, Public Access, Logs, and Keys

Project & Development

Cloud IaaS Misconfiguration Review: Lock Down Permissions, Public Access, Logs, and Keys

Advanced | 4 whiteboard pages · Task board: 5 columns (4 parent tasks, 12 subtasks)

Advanced: Assumes you know the technical terms and methods, and the steps are only outlined. Work through it on your own, guided by decision criteria and points to check.

Teams running systems on AWS, Azure, or Google Cloud use this cloud misconfiguration review template to check permissions, public settings, network entry points, logs, and keys. You check all five areas in the console, settings, and change records (no probing from outside), fix what you find, and use organization policies and automated checks so mistakes do not return. First, split what the provider secures from what you secure under the shared responsibility model; the split varies by service, so check each provider's latest documentation. Four common pitfalls each get what happens, how to check, and how to prevent it: strong permissions left after an incident, public test storage, no log destination or detailed logging (untraceable after default retention ends), and daily use of the top admin account. Changes follow four steps, and every exception gets an expiry and an approver. The whiteboard has 4 pages; the task board has 5 columns, 4 parent tasks, and 12 subtasks.

0 Likes 0 Copies Author: WeProcess
cloud security misconfiguration access management shared responsibility AWS
Sign up free and use this template

Once you have a free account, you can copy this template from the marketplace

Whiteboard preview

Read the whiteboard text as plain text

Cloud IaaS Configuration Review

How to use (4 pages): ① Page 1: set the goal and responsibility scope, list environments and admins (boards①–③) → ② Page 2: inspect the 5 areas and check the pitfalls (boards④–⑦) → ③ Page 3: rank what you found and set the fix order and dates (boards⑧–⑨) → ④ Page 4: set guardrails and the 4 change steps, then keep checking (boards⑩–⑭)

1/4 Set who secures what

① Review goal and scope (start writing here)

  • Review goal e.g. On the production environment holding customer data, remove readable-by-anyone settings and overly strong permissions
  • Target clouds and environments e.g. Two AWS accounts (production and test) and one Google Cloud project for analytics
  • Period and team e.g. Two weeks. The infra lead reviews and one more developer checks

② Shared responsibility model (split what the provider and you secure)

What the provider secures
  • e.g. Data center buildings, power, and physical servers
  • e.g. The virtualization layer and operation of the services themselves
  • The split varies by service type. Confirm it in each provider's latest documentation
What you secure
  • e.g. Account, permission, and multi-factor authentication settings
  • e.g. Data public settings, inbound network rules, logs, and keys
  • e.g. OS and middleware updates on virtual machines (for IaaS)

③ Admins for each environment (pitfall: using the top admin account every day)

Account name and purpose Admins (primary and backup) Handling the top admin account (what happens, how to check, how to prevent)
Production e.g. AWS production account for the customer-facing service e.g. Primary: Infra lead A, backup: Dev lead B e.g. What happens: used daily, one slip or takeover affects everything
Test e.g. AWS test account for pre-release checks e.g. Primary: Dev lead B, backup: Infra lead A e.g. How to check: see the last-use date in the sign-in history and whether MFA is on
Development e.g. Google Cloud project for analytics e.g. Primary: Data lead C, backup: ____ e.g. Prevention: work day to day as a narrowly scoped personal user, and log each use in the register

2/4 Inspect the settings

④ Check the 5 areas for each environment in ③

How to check (screen or record) Passing standard Result Owner
Permissions e.g. In the user and role list, see who holds admin rights and their last-used dates e.g. At most two admins. MFA for everyone (SMS beats nothing but is weaker than authenticator apps or passkeys) e.g. Action needed: one former employee's user remains e.g. Infra lead A
Public settings e.g. In the storage and database settings, read the public settings and access policies e.g. Nothing is readable by anyone. If an account-wide public-access block exists, it is on e.g. Action needed: test storage is still public e.g. Dev lead B
Network entry points e.g. In the inbound rules settings, look for management ports open to every source e.g. Management ports are reachable only from the office or an approved route e.g. No issue e.g. Infra lead A
Logs e.g. In the audit log settings, see which scope is enabled and where logs go e.g. Every account and region sends management activity to a long-term store, and the retention period is written down (default retention differs by service, so check each provider's guidance) e.g. Action needed: the test account has no long-term logging e.g. Infra lead A
Keys and secrets e.g. In the key list, see long-lived keys and last-used dates. Match secret locations against the register e.g. Stop unused keys. Keep secrets only in the approved vault e.g. Action needed: two long-unused keys e.g. Data lead C

⑤ Pitfall: forgetting to remove strong work permissions

  • What happens: e.g. An admin permission granted during an incident stays and magnifies the damage of a slip or takeover
  • How to check: e.g. In the permission list, see the grant date, last-used date, and the recorded reason
  • How to prevent: e.g. Grant strong permissions with an expiry and record the removal in the ⑪ record at the end of the work
  • Result: e.g. One admin permission granted for an incident was still there (⑧ sets the fix order)

⑥ Pitfall: it is only a test environment, so it stays public

  • What happens: e.g. Test storage holding a copy of production data becomes readable by anyone
  • How to check: e.g. In the test environment's settings, see the public settings and match the data types stored against the register
  • How to prevent: e.g. Apply the same prohibition policy to test and never copy production data
  • Result: e.g. One test storage is still public and held a copy of production customer data (to ⑧)

⑦ Pitfall: logs are not enabled

  • What happens: e.g. Past the default retention (it differs by provider, so check the latest guidance), you cannot trace who changed which setting and when
  • How to check: e.g. Match against the list in ③ and see that audit logs are kept long term and data access logging is on, even in new environments and rarely used regions, and that there is no record of logging being stopped
  • How to prevent: e.g. Put turning on logging first in the runbook for creating a new environment
  • Result: e.g. The test account had no long-term logging and no storage destination was set (to ⑧)

3/4 Decide the fix order

⑧ Rank what ④–⑦ found by impact and ease of fixing

Vertical axis: Impact (top = large / bottom = small) / Horizontal axis: Ease of fixing (left = takes effort / right = quick to fix)

  • Plan the fix High impact but takes effort. Set the owner and deadline in ⑨ and write a temporary measure e.g. Split the production admin rights again by role
  • Fix this week High impact and quick to fix. Put it in the top rows of ⑨ e.g. Make the test storage private and disable the former employee's user
  • Revisit at next review Low impact and takes effort. Add it to the scope of the periodic review in ⑬ e.g. Clean up old development environments
  • Fix along the way Low impact and quick to fix. Clear it together with other work e.g. Stop the two unused keys

⑨ From the upper right of ⑧, decide what to fix, who owns it, and by when

What to fix Owner (who fixes, who checks) Deadline
Priority 1 e.g. Make the test storage private e.g. Dev lead B, checked by Infra lead A e.g. This Friday
Priority 2 e.g. Send the test account's audit logs to a long-term store e.g. Infra lead A, checked by Dev lead B e.g. This Friday
Priority 3 e.g. Disable the former employee's user and remove strong permissions e.g. Infra lead A, checked by Data lead C e.g. This Friday
Priority 4 e.g. Split the production admin rights again by role e.g. Infra lead A, checked by Dev lead B e.g. End of week 2

4/4 Prevent repeat mistakes

⑩ Guardrails and exceptions so what ⑨ fixed does not return

Decided by organization policy Exception conditions Exception approver and deadline
No public access e.g. Never make storage or databases readable by anyone e.g. Only items meant to be public, such as an image bucket for a public site e.g. Dev lead B, reviewed every six months
Strong permissions e.g. Grant strong permissions with an expiry and remove them when the work ends e.g. Only during incident response, with the reason recorded e.g. Infra lead A, until the work ends
Required logging e.g. Send audit logs from every account to a long-term store and prohibit the action that turns this off e.g. No exceptions e.g. None
Keys and secrets e.g. Keep secrets out of code and config files and in the approved vault e.g. Only old systems mid-migration, with a move date set e.g. Data lead C, until the move date

⑪ Four steps for changing a setting

  1. Request Write what, why, and until when you will change it e.g. I want to change the public setting of test storage. Write the reason and the date to revert
  2. Second-person review Someone other than the requester checks it against the ⑩ policy e.g. If the change goes against policy, see whether it meets the exception conditions
  3. Rollout Roll out from the configuration record (code) wherever possible e.g. If you changed it by hand, write down the screen and the value
  4. Record and review Write what changed, the date, and the revert date in the register e.g. On the revert date, check in the ⑬ periodic review that nothing was left behind

⑫ Automated configuration checks

  • e.g. A tool that reads each provider's settings checks every day for public settings and for audit logs being on
  • e.g. If it finds a setting that breaks the ⑩ policy, the owner fixes it using the four steps in ⑪
  • Owner: ____

⑬ Periodic review

  • e.g. Every three months, re-check the five areas of ④ with the same methods
  • e.g. Also look at exception deadlines and the records of strong permissions granted and removed
  • Next review date: e.g. First Monday three months from now

⑭ Alerts (set on the cloud side)

  • e.g. Notify the owner when someone signs in with the top admin account
  • e.g. Notify the owner at once if the audit log setting (sending to the long-term store) is stopped or changed
  • Who gets alerts: e.g. Infra lead A and Dev lead B

Task board contents

Not started 16 items

Set responsibility

On page 1, write the goal and scope, confirm which part the shared responsibility model leaves to you, and list the admins for each environment. Fixing the scope here makes it clear on page 2 what to inspect in which environment. Press Ctrl+F (⌘F on Mac) and type "3/4" to jump to page 3.

List your accounts

In each provider's console, open the list of accounts you use (AWS accounts, Azure subscriptions, Google Cloud projects) and copy names and purposes into the rows of board ③. It takes about 15 minutes and is done when the list is filled.

Confirm your share

For each type of service you use, read the provider's latest shared responsibility documentation and write the provider's scope and your own in the two columns of board ②. Done when your column names specific settings.

Write goal and team

On board ①, write the goal, scope, period, and team, then add primary and backup admins and how the top admin account is handled for each environment in ③. Done when the reviewer and the checker are different people and ③ has no blanks.

Inspect settings

On page 2, inspect the five areas for each environment in ③ using the console and settings records, and check the three pitfalls. This is the core parent task and takes the most time.

Fill the 5 checks

In the table on board ④, look at permissions, public settings, network entry points, logs, and keys as the check column says, and write the result and owner. Done when the result column has no blanks.

Find leftover admin

For the pitfall on board ⑤, check the permission list and last-used dates for strong permissions granted for work and never removed, then write what happens, how to check, how to prevent it, and the result. Done when the leftovers are written so they can pass to ⑧.

Check public & logs

For the pitfalls on boards ⑥ and ⑦, check the test environment's public settings and whether logging is enabled in the settings screens, and write what happens, how to check, how to prevent it, and the result. Done when both frames have results.

Set the fix order

On page 3, rank what ④ to ⑦ found by impact and ease of fixing, then decide what to fix, who owns it, and by when. This parent task lets you start with the highest impact.

List what you found

On the four quadrants of board ⑧, place each item from ④ to ⑦ that needs action, one per card. Done when every item sits in a quadrant.

Set order and dates

In the table on board ⑨, write what to fix, the owner, and the deadline, starting from the upper right of ⑧. Write deadlines like "this Friday" or "week 2".

Verify the fixes

Have someone other than the reviewer re-check what ⑨ fixed, using the same method as ④. When the result is no issue, move the card on from the Peer check column.

Build guardrails

On page 4, decide the guardrails and exceptions that stop fixed misconfigurations from returning, the four steps for changing settings, and the ongoing checks. Track them in the Guardrail in place column of the task board.

Set guardrails

In the table on board ⑩, write what organization policy prohibits, the exception conditions, and each exception's approver and deadline. Done when every exception has a deadline.

Define the 4 steps

On the four steps of board ⑪, write how to request, have another person review, roll out, and record and review, then agree on it as a team. Done when all four steps say who does what.

Plan ongoing checks

On boards ⑫ to ⑭, write the content and owner of automated configuration checks, periodic reviews, and the alerts set on the cloud side. Done when the next review date is set.

Reviewing 0 items

Fixing 0 items

Peer check 0 items

Guardrail in place 0 items

← Back to templates

Preview mode Cloud IaaS Misconfiguration Review: Lock Down Permissions, Public Access, Logs, and Keys

Whiteboard preview

① Review goal and scope (start writin... ③ Admins for each environment (pitfal... ④ Check the 5 areas for each environm... ⑤ Pitfall: forgetting to remove stron... ⑥ Pitfall: it is only a test environm... ⑦ Pitfall: logs are not enabled ⑧ Rank what ④–⑦ found by impact and e... Plan the fix Fix this week Revisit at next review Fix along the way ⑨ From the upper right of ⑧, decide w... ⑩ Guardrails and exceptions so what ⑨... ⑪ Four steps for changing a setting ⑫ Automated configuration checks ⑬ Periodic review ⑭ Alerts (set on the cloud side) Cloud IaaS Config... How to use (4 pag... Goal and what to ... Decisions and how... Pitfalls and prob... Prevention and me... 1/4 Set who secures what Ctrl+F / ⌘F : 2/4... → 2/4 2/4 Inspect the settings ↙ 3/4 Review goale.g. On the production environment holding customer data, remove readable-by-anyone settings and overly strong permissions Target clouds and environmentse.g. Two AWS accounts (production and test) and one Google Cloud project for analytics Period and teame.g. Two weeks. The infra lead reviews and one more developer checks ② Shared responsi... What the provider secures e.g. Data center buildings, power, and physical servers e.g. The virtualization layer and operation of the services themselves The split varies by service type. Confirm it in each provider's latest documentation What you secure e.g. Account, permission, and multi-factor authentication settings e.g. Data public settings, inbound network rules, logs, and keys e.g. OS and middleware updates on virtual machines (for IaaS) Account name and ... Admins (primary a... Handling the top ... Production e.g. AWS production account for the customer-facing service e.g. Primary: Infra lead A, backup: Dev lead B e.g. What happens: used daily, one slip or takeover affects everything Test e.g. AWS test account for pre-release checks e.g. Primary: Dev lead B, backup: Infra lead A e.g. How to check: see the last-use date in the sign-in history and whether MFA is on Development e.g. Google Cloud project for analytics e.g. Primary: Data lead C, backup: ____ e.g. Prevention: work day to day as a narrowly scoped personal user, and log each use in the register How to check (scr... Passing standard Result Owner Permissions e.g. In the user and role list, see who holds admin rights and their last-used dates e.g. At most two admins. MFA for everyone (SMS beats nothing but is weaker than authenticator apps or passkeys) e.g. Action needed: one former employee's user remains e.g. Infra lead A Public settings e.g. In the storage and database settings, read the public settings and access policies e.g. Nothing is readable by anyone. If an account-wide public-access block exists, it is on e.g. Action needed: test storage is still public e.g. Dev lead B Network entry points e.g. In the inbound rules settings, look for management ports open to every source e.g. Management ports are reachable only from the office or an approved route e.g. No issue e.g. Infra lead A Logs e.g. In the audit log settings, see which scope is enabled and where logs go e.g. Every account and region sends management activity to a long-term store, and the retention period is written down (default retention differs by service, so check each provider's guidance) e.g. Action needed: the test account has no long-term logging e.g. Infra lead A Keys and secrets e.g. In the key list, see long-lived keys and last-used dates. Match secret locations against the register e.g. Stop unused keys. Keep secrets only in the approved vault e.g. Action needed: two long-unused keys e.g. Data lead C What happens: e.g. An admin permission granted during an incident stays and magnifies the damage of a slip or takeover How to check: e.g. In the permission list, see the grant date, last-used date, and the recorded reason How to prevent: e.g. Grant strong permissions with an expiry and record the removal in the ⑪ record at the end of the work Result: e.g. One admin permission granted for an incident was still there (⑧ sets the fix order) What happens: e.g. Test storage holding a copy of production data becomes readable by anyone How to check: e.g. In the test environment's settings, see the public settings and match the data types stored against the register How to prevent: e.g. Apply the same prohibition policy to test and never copy production data Result: e.g. One test storage is still public and held a copy of production customer data (to ⑧) What happens: e.g. Past the default retention (it differs by provider, so check the latest guidance), you cannot trace who changed which setting and when How to check: e.g. Match against the list in ③ and see that audit logs are kept long term and data access logging is on, even in new environments and rarely used regions, and that there… How to prevent: e.g. Put turning on logging first in the runbook for creating a new environment Result: e.g. The test account had no long-term logging and no storage destination was set (to ⑧) 3/4 Decide the fix order → 4/4 4/4 Prevent repeat mi... Vertical axis: Im... High impact but takes effort. Set the owner and deadline in ⑨ and write a temporary measuree.g. Split the production admin rights again by role High impact and quick to fix. Put it in the top rows of ⑨e.g. Make the test storage private and disable the former employee's user Low impact and takes effort. Add it to the scope of the periodic review in ⑬e.g. Clean up old development environments Low impact and quick to fix. Clear it together with other worke.g. Stop the two unused keys What to fix Owner (who fixes,... Deadline Priority 1 e.g. Make the test storage private e.g. Dev lead B, checked by Infra lead A e.g. This Friday Priority 2 e.g. Send the test account's audit logs to a long-term store e.g. Infra lead A, checked by Dev lead B e.g. This Friday Priority 3 e.g. Disable the former employee's user and remove strong permissions e.g. Infra lead A, checked by Data lead C e.g. This Friday Priority 4 e.g. Split the production admin rights again by role e.g. Infra lead A, checked by Dev lead B e.g. End of week 2 Decided by organi... Exception conditions Exception approve... No public access e.g. Never make storage or databases readable by anyone e.g. Only items meant to be public, such as an image bucket for a public site e.g. Dev lead B, reviewed every six months Strong permissions e.g. Grant strong permissions with an expiry and remove them when the work ends e.g. Only during incident response, with the reason recorded e.g. Infra lead A, until the work ends Required logging e.g. Send audit logs from every account to a long-term store and prohibit the action that turns this off e.g. No exceptions e.g. None Keys and secrets e.g. Keep secrets out of code and config files and in the approved vault e.g. Only old systems mid-migration, with a move date set e.g. Data lead C, until the move date RequestWrite what, why, and until when you will change ite.g. I want to change the public setting of test storage. Write the reason and the date to revert Second-person reviewSomeone other than the requester checks it against the ⑩ policye.g. If the change goes against policy, see whether it meets the exception conditions RolloutRoll out from the configuration record (code) wherever possiblee.g. If you changed it by hand, write down the screen and the value Record and reviewWrite what changed, the date, and the revert date in the registere.g. On the revert date, check in the ⑬ periodic review that nothing was left behind e.g. A tool that reads each provider's settings checks every day for public settings and for audit logs being on e.g. If it finds a setting that breaks the ⑩ policy, the owner fixes it using the four steps in ⑪ Owner: ____ e.g. Every three months, re-check the five areas of ④ with the same methods e.g. Also look at exception deadlines and the records of strong permissions granted and removed Next review date: e.g. First Monday three months from now e.g. Notify the owner when someone signs in with the top admin account e.g. Notify the owner at once if the audit log setting (sending to the long-term store) is stopped or changed Who gets alerts: e.g. Infra lead A and Dev lead B → 2/4 Inspect the settings ↙ 3/4 Decide the fix order → 4/4 Prevent repeat mi...
100%

Drag to move, scroll to zoom