Huawei H13-511 Practice Exam

• A. IaaS Correct answer
• B. PaaS
• C. DaaS
• D. SaaS

Option A, IaaS, is correct because the provider supplies virtual machines, storage and networks, and the customer installs and manages the operating system and applications. PaaS (option B) also supplies the runtime platform, SaaS (option D) delivers a finished application, and DaaS (option C) delivers virtual desktops.

• A. VMs on one host share a single guest OS kernel
• B. A fault in one VM does not affect the other VMs Correct answer
• C. Several VMs run at the same time on one physical server Correct answer
• D. A VM is bound to the CPU model of its original host
• E. A VM is saved as a set of files that can be moved Correct answer

Options B, C and E are correct because virtualization isolates VMs from each other, lets many VMs run simultaneously on one server, and encapsulates each VM as files that can be copied or migrated. Option A is wrong because each VM runs its own guest OS kernel, and option D is wrong because VMs are hardware-independent and not bound to the original host's CPU model.

• A. RAID 5 needs at least four member disks to be built
• B. RAID 0 stores a parity block with every stripe
• C. RAID 1 writes identical data to two member disks Correct answer
• D. RAID 6 tolerates two member disks failing at once Correct answer

Options C and D are correct. RAID 1 mirrors identical data to two member disks, and RAID 6 keeps two parity blocks per stripe so it survives two simultaneous disk failures. Option A is wrong because RAID 5 needs only three disks, and option B is wrong because RAID 0 stripes data with no parity at all.

• A. Midrange server
• B. Tower server
• C. Rack server
• D. Blade server Correct answer

Option D is correct because blade servers are inserted into a shared chassis that provides common power modules, fans and switch modules. Tower and rack servers are standalone units with their own power and cooling, and midrange server describes a class of system rather than a chassis-based form factor.

• A. RAID 6 Correct answer
• B. RAID 10
• C. RAID 5 Correct answer
• D. RAID 0
• E. RAID 1

Options A and C are correct because RAID 5 and RAID 6 compute parity on every write, which adds a write penalty. RAID 0 has no redundancy, and RAID 1 and RAID 10 use mirroring, so none of them calculate parity.

• A. IP SAN
• B. NAS
• C. DAS Correct answer
• D. FC SAN

Option C is correct because DAS attaches disks directly to the server bus or interface with no storage network. NAS, IP SAN and FC SAN all reach storage across a network.

• A. Class E
• B. Class B
• C. Class C
• D. Class D Correct answer

Option D, Class D, is correct because multicast uses 224.0.0.0 to 239.255.255.255. Classes A, B and C are unicast ranges, and Class E (option A) is reserved for experimental use.

• A. Neighbors are configured manually by the administrator Correct answer
• B. It floods link-state advertisements through an area
• C. It discovers neighbors with multicast Hello packets
• D. It records the AS path of each route to avoid loops Correct answer
• E. It selects routes by hop count

Options A and D are correct because BGP neighbors are configured manually and run over a TCP session, and each route carries an AS path used to detect and avoid loops. Flooding link-state advertisements through an area and using Hello multicast are OSPF behaviors, and BGP selects routes by path attributes, not hop count.

• A. /etc Correct answer
• B. /usr
• C. /opt
• D. /var

Option A is correct because /etc is the standard location for system-wide configuration files on Linux. /usr holds installed programs and libraries, /opt holds optional third-party software, and /var holds variable data such as logs and spools.

• A. True Correct answer
• B. False

True. The Linux kernel is released under the GNU General Public License (GPL), which lets users run, study, modify and redistribute the source code under the same license terms.

• A. True Correct answer
• B. False

True. With memory overcommitment enabled, FusionCompute lets the total memory assigned to VMs exceed the host's physical memory by using techniques such as memory sharing, swapping and ballooning. This raises VM density but can reduce performance under heavy load.

• A. Partitioning
• B. Hardware independence
• C. Isolation
• D. Encapsulation Correct answer

Option D, Encapsulation, is correct because a VM is stored as a set of files (configuration and virtual disks) that can be copied, backed up or moved like ordinary data. Partitioning means dividing host resources among VMs, hardware independence means a VM is decoupled from the physical hardware, and isolation means VMs cannot affect each other.

• A. Bare-metal virtualization Correct answer
• B. Application streaming without a hypervisor
• C. Container-based virtualization sharing one kernel
• D. Hosted virtualization on a general-purpose host OS

Option A, bare-metal virtualization, is correct because the hypervisor installs directly on the physical hardware with no host OS beneath it, as FusionCompute does. Hosted virtualization (option D) runs on top of a host OS, and container-based virtualization (option C) shares one kernel instead of using a hypervisor.

• A. True Correct answer
• B. False

True. The VRM (Virtualization Resource Management) node manages the hosts, VMs, storage and networks of a FusionCompute site from one place and provides the web portal. The CNA hosts run the VMs under its control.

• A. HDC
• B. ITA
• C. CNA Correct answer
• D. VRM

Option C, CNA, is correct because the Computing Node Agent runs on every physical server and virtualizes its CPU, memory and network resources for VMs. VRM (option D) is the management node that controls the CNAs, while HDC and ITA are FusionAccess components, not FusionCompute ones.

• A. WI Correct answer
• B. vAG
• C. ITA
• D. HDC

Option A, WI, is correct because the Web Interface provides the login page where users enter credentials and choose a desktop. vAG (option B) is the access gateway for external users, ITA (option C) is the IT adapter used for management integration, and HDC (option D) manages and brokers desktops rather than presenting the login page.

• A. HDC
• B. vAG Correct answer
• C. ITA
• D. GaussDB

Option B, vAG, is correct because the virtual Access Gateway is normally deployed in the DMZ and proxies desktop protocol traffic from Internet users, so they have no direct path to the internal network. HDC (option A) and ITA (option C) are internal management components, and GaussDB (option D) is the database.

• A. ITA
• B. WI
• C. CNA
• D. HDA Correct answer

Option D, HDA, is correct because the Huawei Desktop Agent installed in the guest OS lets the desktop register with HDC and be managed by it. ITA and WI are server-side FusionAccess components, and CNA is the FusionCompute host agent, none of which belong in a desktop template.

• A. The binding between Li and the desktop is retained Correct answer
• B. The desktop returns to the pool for other users
• C. Li moves to another group automatically
• D. HDC powers off the VM and releases its disks
• E. Li cannot start a new session on that desktop Correct answer

Options A and E are correct because maintenance mode only stops new sign-ins to that desktop; the user binding stays, so the desktop is still Li's. It does not return the desktop to the pool, move Li to another group, or power off the VM and release its disks.

• A. Place the model in the cloud region with the lowest storage price
• B. Stream all camera video to a central cloud region for inference
• C. Run the inference model on an edge node located in the factory Correct answer
• D. Grow the central cluster so the uplink round trip becomes shorter
• E. Send summary results and sample frames to the cloud for retraining Correct answer

Options C and E are correct because a few milliseconds of latency can only be met by running inference next to the cameras on an edge node, and sending only summaries and sample frames keeps the limited uplink usable while still feeding cloud retraining. Streaming all video to the cloud (B) exhausts the bandwidth and adds delay, a cheap-storage region (A) is far from the line, and a bigger central cluster (D) does not shorten the physical round trip.