Free tech learning for everyone, from basics to pro

Network ArchitectureNetworking SecurityIntermediate

Network performance, RAID, load balancing and backups

In short: Performance and availability go together: RAID protects against disk failure, NIC teaming and load balancing remove single points of failure, and a sound backup scheme (full plus differential or incremental) means data can be restored after a disaster.

Measuring network performance

MetricWhat it tells you
BandwidthThe maximum capacity of a link
ThroughputHow much data actually gets through
LatencyHow long data takes to travel from source to destination
JitterHow much latency varies; critical for voice and video
Packet lossThe share of packets that never arrive
AvailabilityThe percentage of time a service is up (for example 99.9%)

RAID storage

RAID (redundant array of independent disks) combines disks to gain speed, redundancy or both. RAID is not a backup: deleting a file or a ransomware attack affects every disk at once.

RAID 0, RAID 1 and RAID 5 block layoutRAID 0: stripingA1A3A5A2A4A6fast, no redundancyRAID 1: mirroringA1A2A3A1A2A3survives one disk failureRAID 5: parityA1B2CpA2BpC1ApB1C2survives one disk failure
Figure 1. Striping spreads blocks for speed, mirroring copies them, and parity (p) lets a lost disk be rebuilt from the others.
  • RAID 0 (striping): data is split across disks for speed and full capacity, but one failed disk loses everything.
  • RAID 1 (mirroring): identical copies on two disks; if one fails the other carries on, and a replacement is rebuilt from it.
  • RAID 5 (striping with parity): at least three disks; parity is spread across all of them, so any one disk can fail.
  • RAID 6 (double parity): like RAID 5 with a second parity block, so two disks can fail at once.
  • RAID 10 (mirroring + striping): at least four disks; mirrored pairs are striped for both speed and redundancy, at the cost of half the raw capacity.

Dual LAN (NIC teaming) and load balancing

NIC teaming joins two or more network cards in a server into one logical link, ideally cabled to different switches. If a card, cable or switch fails, traffic continues on the other, and in active-active mode both add bandwidth.

A load balancer spreads requests across several servers and runs health checks so failed servers stop receiving traffic. Common methods:

  • Round robin: each new request goes to the next server in turn.
  • Least connections: the server with the fewest active sessions gets the next one.
  • Source IP hash: a client’s address always maps to the same server, keeping sessions together.

Asset management

You cannot protect what you do not know you have. Asset management keeps an up-to-date inventory of hardware, software and licences: owner, location, serial number, software version, warranty and support dates. It shows which devices need patching, which are out of support and which have gone missing.

Backup types

Full, differential and incremental backups over four daysMonTueWedThuFullFullFullFullFullDifferentialFullsince Monsince Monsince MonIncrementalFullsince Monsince Tuesince Wed
Figure 2. A differential grows each day until the next full backup; an incremental only holds changes since the previous backup.
TypeBacks upBackup speed and spaceTo restore you need
FullEverythingSlowest, largestThe one full backup
DifferentialChanges since the last fullGrows each dayLast full + latest differential
IncrementalChanges since the last backup of any kindFastest, smallestLast full + every incremental since

The 3-2-1 rule: keep three copies of data, on two different types of media, with one copy off-site or offline. Test restores regularly.

Storage area networks (SAN)

A SAN is a dedicated high-speed network that connects servers to shared block storage. Servers see SAN storage as if it were a local disk. SANs run over Fibre Channel or over Ethernet with iSCSI, and they make it easier to replicate, snapshot and move data between servers and sites.

Practice

Which RAID level survives two simultaneous disk failures?
RAID 6 (double parity). RAID 10 can too if the failed disks are in different mirrored pairs.
A full backup runs on Sunday and incrementals every other night. A disk dies on Thursday. What do you restore?
Sunday’s full, then Monday’s, Tuesday’s and Wednesday’s incrementals in order.
Why is RAID not a substitute for backup?
RAID copies every change instantly, including deletions and encryption by ransomware.

Key takeaways

  • Measure bandwidth, throughput, latency, jitter, packet loss and availability.
  • RAID 0 is speed only; RAID 1, 5, 6 and 10 add redundancy. RAID is not backup.
  • NIC teaming and load balancing remove single points of failure.
  • Differential restores need two sets; incremental restores need the whole chain.
  • A SAN gives servers fast shared block storage.
Tracks your learning on this device only. See your progress on the learning paths.

About the author

Mohamed Fayas

Founder of FM TecHub. CCNA certified, with a BTEC HND in Network Engineering, and BEng (Hons) in Computer Networking and Cloud Security. He writes practical, plain-English guides on networking, security and cloud.

More about the author

Original content written by Mohamed Fayas for FM TecHub. © 2026 FM TecHub. You're welcome to share a link to this page; please don't copy the text elsewhere without permission.

Stay up to date by email

New notes, tools and security news from FM TecHub, explained simply. Free.

We only use your email to send these updates. See our privacy policy.