Metrics
Security Center uses tenant log events to identify patterns that are usually an indicator of known attack types. We classify tenant log event patterns into categories: normal traffic, credential stuffing threats, signup attack threats, and MFA bypass threats.
Classification of event type codes may change. Avoid implementing solutions dependent on the current log event code definitions.
Normal traffic
We use normal traffic to establish a benchmark against different threat types we may observe. Normal traffic includes all successful and failed events for a given hour, which includes the following event codes:
| Event code | Event |
|---|---|
s |
Successful login |
ss |
Successful signup |
f |
Failed user login |
fu |
Failed user login due to invalid username |
fp |
Failed user login due to invalid password |
pwd_leak |
Attempted login with a leaked password |
Credential stuffing
We identify credential stuffing threats within a single hour with the following event codes:
| Event code | Event |
|---|---|
f |
Failed user login |
fu |
Failed user login due to invalid username |
fp |
Failed user login due to invalid password |
fs |
Failed signup |
pwd_leak |
Attempted login with a leaked password |
limit_wc |
IP blocked for >10 failed login attempts to a single account |
limit_sul |
User blocked for >20 login per minute from the same IP address |
limit_mu |
IP blocked for >100 failed login attempts or >50 signup attempts |
Signup attack
We identify signup attack threats within a single hour with the following event codes:
| Event code | Event |
|---|---|
fs |
Failed signup |
MFA bypass
We identify MFA bypass threats within a single hour with the following event codes:
| Event code | Event |
|---|---|
cs |
Sent code |
cls |
Sent code/link |
gd_send_pn |
Sent push notification |
gd_send_sms |
Sent SMS |
gd_auth_failed |
Failed OTP authentication |
gd_auth_rejected |
Rejected OTP authentication |
gd_otp_rate_limit_exceed |
Too many OTP authentication failures |
gd_recovery_failed |
Failed recovery |
gd_recovery_rate_limit_exceed |
Too many recovery failures |