Server Basics · Topic 1 of 1
Servers and Key Server Technologies
Learn server form factors, x86 and ARM architectures, server hardware, RAID 0, 1, 5, 6 and 10, hot spares, and BMC, IPMI and iBMC out-of-band management.
- 5 min
- Form factors: tower, rack, blade and high density; rack height is measured in U.
- x86 is CISC; ARM (Huawei Kunpeng) is RISC.
- RAID 0 striping, RAID 1 mirroring, RAID 5 one parity, RAID 6 two parities, RAID 10 mirrors striped.
What it is
A server is a computer that provides services to other computers over a network. Compared with a PC it is built for reliability, availability and serviceability (RAS), with redundant parts and remote management.
In a cloud data center, servers are the compute layer. FusionCompute CNA hosts are ordinary servers, so their hardware matters to a cloud engineer.
How it works
Server form factors
| Form factor | Description | Typical use |
|---|---|---|
| Tower | Stands alone like a large PC | Small offices with few servers |
| Rack | Fixed height in U (1U = 44.45 mm) in a standard 19-inch rack | Most data center servers |
| Blade | Thin server blades share the chassis power, fans and switching | High density with central management |
| High density | Several server nodes in one rack chassis | Cloud and HPC clusters |
CPU architectures
| Architecture | Instruction set | Examples |
|---|---|---|
| CISC | Complex, variable length instructions | x86 processors from Intel and AMD |
| RISC | Simple, fixed length instructions | ARM, including Huawei Kunpeng; Power and SPARC in UNIX servers |
Server hardware
| Component | Role |
|---|---|
| CPU | Executes instructions; servers often have two or more sockets |
| Memory | Holds running data; server memory uses ECC to detect and correct bit errors |
| Hard disks | Local HDDs or SSDs, usually hot swappable |
| RAID controller card | Builds RAID arrays from the disks and caches I/O |
| NIC | Network ports, often separate ones for service, storage and management traffic |
| Power supply units | Usually two in 1+1 redundancy, hot swappable |
| Fans | Cooling, redundant and hot swappable |
| Mainboard | Connects the components through buses such as PCIe |
Hot swap means replacing a part while the server keeps running.
Key technology 1: RAID
RAID (redundant array of independent disks) combines several disks into one logical disk for more capacity, speed or protection. It uses three methods: striping spreads data across disks, mirroring writes the same data to two disks, and parity stores XOR check data that can rebuild a lost disk.
| Level | Method | Minimum disks | Usable capacity (N equal disks) | Disk failures survived |
|---|---|---|---|---|
| RAID 0 | Striping only | 2 | N | None |
| RAID 1 | Mirroring | 2 | N/2 | One per mirror pair |
| RAID 5 | Striping with distributed parity | 3 | N minus 1 | One |
| RAID 6 | Striping with two independent parities | 4 | N minus 2 | Two |
| RAID 10 | Mirror pairs, then striped (RAID 1+0) | 4 | N/2 | One per mirror pair |
RAID 0 is fastest but loses everything if one disk fails. RAID 6 adds a second parity for large disks with long rebuilds.
A hot spare disk stands by in the array. When a member fails, the controller automatically rebuilds the lost data onto the hot spare. During the rebuild performance drops, and in RAID 5 a second failure before it completes loses the data.
Key technology 2: BMC and IPMI
The BMC (baseboard management controller) is a small independent processor on the mainboard with its own management network port. It runs on standby power, so it works when the server OS has crashed or the server is powered off.
IPMI (Intelligent Platform Management Interface) is the open standard for managing servers through the BMC, first published in 1998. Huawei servers use iBMC, Huawei's own BMC, which supports IPMI and newer interfaces such as Redfish.
The BMC provides:
- Monitoring of temperature, voltage, fans and power supplies, with alarms and logs
- Remote power on, power off and reset
- A remote console (KVM over IP) and virtual media to install an OS without visiting the server
This is out-of-band management: it uses a separate management path, independent of the service network and the OS. In-band management goes through the running OS, for example an SSH session to the server.
A common open source client is ipmitool:
ipmitool -I lanplus -H 192.168.10.20 -U <user> -P <password> chassis power statusA worked example
A server has six 1.2 TB disks. RAID 5 gives 6 TB and survives one failed disk, and RAID 6 gives 4.8 TB and survives two. RAID 10 gives 3.6 TB and survives one failure per mirror pair.
One night the host OS hangs. The administrator opens the iBMC web page, uses the remote console and power cycles the server without visiting the data center.
How the exam asks about this
- "Which RAID level offers no redundancy?" → RAID 0.
- "How many disks can fail in a RAID 6 array without data loss?" → Two.
- "Which component lets an administrator power on a server when the OS is not running?" → The BMC (iBMC on Huawei servers), through out-of-band management.
What to remember for the exam
- Form factors: tower, rack, blade and high density; rack height is measured in U.
- x86 is CISC; ARM (Huawei Kunpeng) is RISC.
- RAID 0 striping, RAID 1 mirroring, RAID 5 one parity, RAID 6 two parities, RAID 10 mirrors striped.
- A hot spare replaces a failed disk automatically, and the controller rebuilds onto it.
- The BMC gives out-of-band management through IPMI; Huawei's version is iBMC.
RAID is not a backup. It protects against disk failure, but deleted files, corruption and ransomware are copied to every disk at once.
Common mistakes
Counting RAID 5 capacity wrongly. Parity is spread across all disks, not kept on one disk, but its total size still equals one disk. Usable space is N minus 1 disks.
Treating RAID 10 as RAID 0 plus spare disks. RAID 10 mirrors first and then stripes the mirror pairs. It loses half the raw capacity and survives one failure per pair.
Thinking the BMC needs the OS. The BMC is separate hardware on standby power. It works when the OS is down and even when the server is switched off.
Check yourself
What is the usable capacity of four 2 TB disks in RAID 10?
4 TB. RAID 10 mirrors the disks in pairs, so half of the 8 TB raw capacity is usable.
What is the minimum number of disks for RAID 5?
Three. RAID 5 stripes data with one disk's worth of distributed parity.
What is the difference between in-band and out-of-band management?
In-band management goes through the running OS, such as SSH. Out-of-band management goes through the BMC's own management port and works even when the OS is down.
Topic 1 of 1 in Server Basics
Test yourself on Server Basics
The full Huawei H13-511 question bank has 600 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Huawei. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.