The ports and protocols every SOC analyst needs to know -- 22, 25, 53, 80, 443, 445, 3389, 1433, 3306, 5985/5986 and more -- plus what traffic on each port means during an investigation.
The Ports Every Analyst Must Know and What They Mean in Context
Port numbers identify specific services on a host. The first 1024 ports are well-known ports assigned by IANA. When an analyst sees a network connection in a SIEM alert, the destination port tells them which service was being accessed — which immediately narrows what the attacker was doing. The source port is usually ephemeral (1024-65535) and rarely meaningful.
The table below covers every port you’ll encounter in a workday SOC investigation. For each port, the table covers: what the service normally does, what it looks like in logs, and the suspicious patterns that warrant escalation.
Web and Email Services
Port
Protocol
Service
Normal Use
Suspicious Indicators
80
TCP
HTTP
Standard web traffic. Web servers, API endpoints, redirect targets.
Unencrypted traffic in a HTTPS-first environment suggests a downgrade attack or misconfigured internal app. HTTP traffic carrying binary payloads or large POST bodies may indicate data exfiltration or C2.
Nearly everything runs on 443. Suspicious patterns: connections to known-bad IPs, unusual JA3/JA3S fingerprints, connections to domains registered < 30 days ago, unusually long-lived sessions.
8080
TCP
HTTP-Alt
Alternative HTTP port. Common for development servers, proxies, application servers.
Outbound connections on 8080 from production servers. Often abused by malware as a fallback C2 port when 443 is blocked.
8443
TCP
HTTPS-Alt
Alternative HTTPS port. Common for admin interfaces, API gateways.
Admin interfaces exposed on non-standard ports. Check if the destination belongs in your asset inventory.
25
TCP
SMTP
Email submission between mail servers (MTA-to-MTA).
Open relay detection: any inbound SMTP connection on 25 from the internet that isn’t from a known mail exchange. Outbound 25 from non-mail servers indicates spam bot activity.
465
TCP
SMTPS
SMTP over SSL/TLS (deprecated but still in use).
Same as port 25. Increasingly rare — flag connections for review.
587
TCP
SMTPS
SMTP with STARTTLS (modern mail submission).
Legitimate mail clients use this to send mail through authenticated relay. Abuse: compromised credentials used to relay spam.
110
TCP
POP3
Post Office Protocol — downloading email from server to client.
Very rare in modern environments (most orgs use IMAP or Outlook/Exchange). Any POP3 traffic after ~2020 is suspect — likely a legacy app or an attacker exfiltrating email archives.
143
TCP
IMAP
Internet Message Access Protocol — managing email on server.
Normal for configured email clients. Suspicious: a client downloading large volumes of emails in bulk (mailbox enumeration or data theft).
993
TCP
IMAPS
IMAP over SSL/TLS.
Same as port 143. Monitor for bulk download patterns.
995
TCP
POP3S
POP3 over SSL/TLS.
As rare as port 110. Heavy traffic on 995 warrants investigation.
Remote Administration
Port
Protocol
Service
Normal Use
Suspicious Indicators
22
TCP
SSH
Secure shell — remote administration, file transfer (SCP/SFTP), tunneling.
Outbound SSH from a server to the internet: almost always suspicious (data exfiltration, C2 tunnel, or pivot). Inbound SSH from unexpected source IPs (especially from countries where your org has no presence). Failed auth volume > 10/min suggests brute forcing.
23
TCP
Telnet
Unencrypted remote terminal access.
Should not exist in any modern network. Any Telnet traffic is a finding. Credentials are sent in plaintext — packet capture reveals them immediately.
3389
TCP
RDP
Remote Desktop Protocol — Windows remote GUI access.
The most attacked remote access port. External RDP brute force attempts are constant. Monitor for: RDP from unexpected source IPs, off-hours RDP sessions, RDP from admin workstations to multiple servers (lateral movement, T1021.001).
Similar to RDP but less monitored. Any external VNC connection is extremely suspicious.
5985
TCP
WinRM-HTTP
Windows Remote Management (HTTP).
Used for PowerShell remoting and system management. Suspicious: WinRM from a non-admin workstation to a domain controller — lateral movement indicator (T1021.006).
5986
TCP
WinRMS
Windows Remote Management (HTTPS).
Same as 5985 but encrypted. Monitor for atypical source-destination pairs.
File and Print Services
Port
Protocol
Service
Normal Use
Suspicious Indicators
445
TCP
SMB
Server Message Block — file sharing, printers, named pipes.
External SMB (port 445 from internet) = immediate incident. Internally: EternalBlue/MS17-010 exploits targeted 445. Abnormal SMB connections between segmented zones suggest lateral movement. Large file transfers to an unusual destination suggest data exfiltration.
139
TCP
NetBIOS-SSN
NetBIOS Session Service (legacy file/print sharing).
Increasingly rare. Predecessor to SMB over 445. Traffic on 139 in a modern Windows domain is unusual — check if legacy apps require it.
137-138
UDP
NetBIOS-NS/DGM
NetBIOS Name Service / Datagram Service.
Name resolution broadcasts. Excessive NetBIOS traffic + SMB on 445 often accompanies Worm-like spreading (e.g., WannaCry malware patterns).
2049
TCP/UDP
NFS
Network File System (UNIX/Linux file sharing).
Unusual in Windows-centric environments. Monitor for: NFS mounts from unexpected sources, large file transfers outside business hours.
Authentication and Directory Services
Port
Protocol
Service
Normal Use
Suspicious Indicators
88
TCP/UDP
Kerberos
Active Directory authentication. Ticket requests and renewals.
Kerberos from a non-domain-joined machine = suspicious. AS-REP roasting detection: high volume of AS-REQ without pre-authentication (T1558.004). Kerberoasting: many TGS-REQ for service tickets (T1558.003).
LDAP queries from non-admin accounts enumerating users/groups = reconnaissance (T1482). High query volume suggests directory enumeration for password spray targets.
636
TCP
LDAPS
LDAP over SSL/TLS.
Same as 389 but encrypted. Any unauthenticated LDAP query from a new source is suspicious.
3268-3269
TCP
Global Catalog
Global Catalog for large Active Directory forests (plaintext/TLS).
GC queries from non-domain controllers are unusual — investigate.
Database Ports
Port
Protocol
Service
Normal Use
Suspicious Indicators
1433
TCP
MSSQL
Microsoft SQL Server.
Outbound 1433 from an application server is normal. Inbound 1433 from the internet is almost never legitimate. Brute force attempts against SQL Server are common. Monitor for large query results (SELECT *) indicating bulk data theft.
3306
TCP
MySQL
MySQL/MariaDB database.
Same pattern as 1433. Inbound from internet = bad. Outbound from unexpected app servers = potential pivot point.
5432
TCP
PostgreSQL
PostgreSQL database.
Same pattern as other databases. Check PgBouncer connections if pooling is in use.
6379
TCP
Redis
Redis key-value store.
Often exposed without authentication. Scan for unauthorized SET commands or data extraction.
27017
TCP
MongoDB
MongoDB database.
Historically subject to mass ransom attacks on exposed instances. Any external MongoDB connection should escalate.
1521
TCP
Oracle DB
Oracle database listener.
Legacy. External connections to 1521 are almost always attackers scanning for exposed Oracle databases.
Infrastructure and Management
Port
Protocol
Service
Normal Use
Suspicious Indicators
53
TCP/UDP
DNS
Domain Name System — name resolution.
UDP 53 for queries, TCP 53 for zone transfers (>512 bytes). Suspicious: DNS TXT query volumes (DNS tunneling indicator, T1572), ANY queries (amplification vector), query volumes > 1000/min to a single domain (beaconing), unusually long subdomain names (data encoding). Zone transfer requests (AXFR) from unauthorized sources.
67-68
UDP
DHCP
Dynamic Host Configuration Protocol.
Rogue DHCP servers handing out malicious DNS/WINS settings. DHCP starvation attacks. Monitor for MAC-IP binding changes.
69
UDP
TFTP
Trivial FTP (no authentication).
Rare in production. Any TFTP traffic is unusual — may indicate firmware manipulation or malware staging.
161
UDP
SNMP
Simple Network Management Protocol (query).
SNMP community string brute forcing. SNMP walking (snmpwalk) leaks device configuration. SNMPv1/v2c sends community strings in plaintext.
162
UDP
SNMP-Trap
SNMP trap receiver.
Unexpected trap sources. Traps from decommissioned devices.
514
UDP
Syslog
System logging (plaintext).
Syslog injection — crafting fake log entries to confuse analysts or hide attacker activity.
123
UDP
NTP
Network Time Protocol.
NTP amplification attacks (monlist command). Check for unexpected monlist responses — your NTP server may be abused as a DDoS amplifier.
EternalBlue (MS17-010) exploited RPC + SMB. RPC traffic between workstation zones and server zones suggests lateral movement. High RPC traffic volume = enumeration (T1046).
445
TCP
SMB
SMB named pipes over RPC.
Named pipe abuse for lateral movement via PsExec, WMI, scheduled tasks.
636
TCP
LDAPS
LDAP over SSL.
See directory services above.
1434
UDP
MSSQL-Monitor
SQL Server Browser/Monitor.
SQL Server instance discovery. Network scans often hit this port.
1701
UDP
L2TP
Layer 2 Tunneling Protocol.
VPN connections. Similar ISAKMP concerns.
1723
TCP
PPTP
Point-to-Point Tunneling Protocol.
Legacy VPN protocol. Deprecated in favor of IPsec/WireGuard. Any PPTP traffic should be reviewed.
1812-1813
UDP
RADIUS
Authentication/Accounting.
Widespread Wi-Fi and VPN authentication. RADIUS compromise means network access control is broken.
5060-5061
TCP/UDP
SIP
VoIP session initiation.
War dialing, toll fraud, SIP scanning. Call data record anomalies.
5353
UDP
mDNS
Multicast DNS (ZeroConf).
Rogue devices announcing services. Useful for identifying unauthorized network devices.
Well-Known Port Ranges for Scanning Context
0-1023: System/well-known ports. Most registered to long-established services. Traffic on an unassigned port in this range is anomalous.
1024-49151: Registered/user ports. Ephemeral on many systems, but also registered to specific apps.
49152-65535: Dynamic/private ports. True ephemeral range on modern Windows (Vista+). Any service listening on this range should be investigated.
Practical Triage Using Destination Port
When an alert fires and you see a destination port, here is how to use it:
Is it a standard port for the destination? If the asset is a web server, expect 80/443. If it’s a database server, expect 1433/3306/5432. Mismatched port + asset type = red flag.
Is the direction unusual? Outbound SSH (22) from a server to the internet is never normal. Inbound 445 from outside the org is never normal.
Is the port a known abuse vector? 445, 3389, 22, 1433 are the most attacked ports. Alert on them hitting unexpected assets.
What is the traffic volume? One connection on port 22 to an IP that resolves to a known cloud provider may be a dev committing code. 100 connections suggests a scan or brute force.