Skip to content
cyber journey

Resource Cheatsheet

A+ Core 1 2.3: Network Services

Professor Messer A+ 220-1201 objective 2.3 — DNS, DHCP, file/print sharing, email, syslog/SIEM, web, AAA, database, NTP, spam gateway, UTM, load balancer, proxy, SCADA/ICS, legacy, embedded and IoT.

Updated
Reading time
4 min read
Detail
A+ Core 1
Detail
Network Services
On this page

In one line: every data centre runs a common set of services — naming, addressing, sharing, security, monitoring and control — and knowing what each does is core troubleshooting knowledge.

Companion to: Professor Messer, CompTIA A+ 220-1201 Core 1, objective 2.3 (network services). The full version is the Network Services class notes; the section overview is the Section 2 sheet.


The services

Service Does Key facts
DNS Name → IP address (and reverse) Distributed, load split by domain, often cached
DHCP Auto-assigns IP config Enterprises run multiple for redundancy
File sharing Central storage SMB (Windows), AFP (macOS) — hidden behind the OS UI
Print service Queues and manages print jobs Protocols: SMB, IPP, LPD
Email server Sends/receives mail Very high uptime expectation
Syslog → SIEM Centralise and correlate logs SIEM = Security Information and Event Manager; needs lots of storage
Web server Serves pages HTTP/HTTPS, pages built in HTML
AAA / authentication server Checks credentials, grants access AAA = Authentication, Authorization, Accounting; redundant (everything depends on it)
Database server Stores relational tables Queried with SQL (MS SQL Server, MySQL…)
NTP Keeps clocks accurate Matters for log correlation and encryption
Spam gateway Filters unwanted email Not 100% accurate
UTM / next-gen firewall / web security gateway All-in-one edge security URL filtering, malware scan, spam filter, firewall, IPS, bandwidth shaping, VPN
Load balancer Spreads traffic across servers Auto-detects and routes around a failed server
Proxy server Requests on the client’s behalf Security, access control, caching, content scanning
SCADA / ICS Remote control of industrial equipment Needs a fully segmented network
Legacy system Old but still running Often still critical — that’s often why it’s still there
Embedded system Purpose-built device No direct OS access; manufacturer supports it
IoT device Networked smart devices Vendors often weak on security → segment its network

Load balancer failover

  1. Requests spread evenly across a server farm.
  2. A server fails → load balancer removes it from rotation automatically.
  3. Remaining servers absorb the load.
  4. Server is fixed → load balancer detects and resumes sending traffic.
  5. Users typically never notice.

Proxy server flow

Client → proxy (on the client’s behalf) → service → response evaluated by proxy → forwarded to client, if safe.


🔐 Security notes

  • Redundant authentication servers are essential — a single point of failure there blocks access to everything.
  • SIEM correlation is how multi-system attacks get noticed — not just storage, a detection tool.
  • Accurate time (NTP) underpins both log investigations and many encryption protocols.
  • A proxy/UTM only protects what’s forced through it — bypasses (e.g. a personal hotspot) undermine the control.
  • Segment what you can’t fully secure directly: SCADA/ICS and IoT share this exact lesson.
  • Legacy and embedded systems often can’t be patched normally — isolate and monitor them.

Practice drills

1. Which service resolves professormesser.com to an IP address, and how is its load spread?

DNS — load is distributed primarily by which domain names a group of servers supports.

2. Why does an enterprise run more than one DHCP server?

Redundancy — if one is unavailable, others still hand out addresses.

3. A user maps a network drive on Windows and shares files on a Mac. Which two protocols are likely at work?

SMB (Windows) and AFP (macOS).

4. What's the difference between syslog and a SIEM?

Syslog is the protocol that sends logs to a central point; the SIEM is the server that consolidates and correlates them.

5. One web server in a farm crashes. What does the load balancer do?

Detects the failure, removes the server from rotation, and spreads the load across the rest — automatically restoring it once it’s healthy again.

6. Why are SCADA/ICS networks usually completely segmented?

They control valuable, often critical physical equipment, so unauthorised remote access must be very hard to achieve.

7. Why should IoT devices go on their own network segment?

Their manufacturers are often not network-security experts, so isolating them limits the blast radius of a compromise.


Key takeaways

  • DNS (name→IP, distributed/cached) and DHCP (auto IP config, redundant) are foundational.
  • File/print sharing: SMB/AFP, IPP/LPD — hidden behind the OS.
  • Email needs high uptime; syslog + SIEM centralise and correlate logs.
  • Web = HTTP/HTTPS + HTML. AAA = auth, authorization, accounting — always redundant.
  • Database servers use relational tables and SQL. NTP keeps time accurate for logs and encryption.
  • Spam gateway, UTM, load balancer, proxy each defend or scale services in a distinct way.
  • SCADA/ICS, legacy, embedded and IoT all share one theme: isolate what’s hard to secure directly.

Search