Runjian Service Desk Interview Experience

Self Introduction

Omitted

Tell me about your project experience

I had just woken up and was too nervous; I couldn’t recall it at the moment… I’m done for!

Which networking protocols are you familiar with

OSPF, BGP, VRRP, stacking

Explain the difference between stacking and M-LAG

Stacking virtualizes multiple devices into one; M-LAG is cross-device link aggregation

Follow-up: Anything else

Stacking merges multiple devices into a single logical whole, with unified management of all ports and a unified control plane M-LAG: each device has an independent control plane, which merely works together to handle the cross-device aggregated links.

Explain the OSPF principle

Link-state routing protocol The main process is: establish neighbors → form adjacencies → flood LSAs to synchronize the LSDB → compute routes locally Simply put, the process is:

1. Each router collects information about its directly connected links and generates LSAs (Link-State Advertisements); instead of passing routing tables directly, it floods the link-state information to all neighbors in the same area
2. After all routers receive all the LSAs, each one stores a copy of the LSDB; the data in the database is identical within the same area. They run the SPF (Shortest Path First) algorithm, computing the shortest path with themselves as root to produce the routing table, selecting the path with the lowest cost
3. Supports area partitioning: backbone area 0, and non-backbone areas, which are further subdivided into: normal areas, stub areas, totally-stub areas, NSSA (Not-So-Stubby Areas), and totally-NSSA
4. Supports equal-cost routes and triggered updates; as soon as a link changes, LSAs are sent, offering fast convergence and multicast interaction

I see Docker on your resume; explain the difference between containers and virtualization?

Virtualization virtualizes the entire hardware layer, where each instance has an independent OS kernel; containers share the kernel ps: I only answered one sentence…

Follow-up: Anything else?

Virtual machines consume more resources but offer stronger isolation; containers consume fewer resources. They are more lightweight but weaker in isolation

Hindsight supplement

1. Virtualization, represented by hypervisors like KVM and VMware, abstracts and virtualizes hardware resources, and each virtual machine has a complete, independent operating system kernel. The Hypervisor layer is responsible for scheduling the underlying hardware, and virtual machines are strongly isolated from each other at the hardware level. Its drawbacks are relatively high resource overhead: images are usually on the order of GB and booting takes minutes; its advantage is strong isolation — different virtual machines can run different operating systems, and a failure in one VM generally does not affect other instances.

2. Containers, such as Docker, are operating-system-level virtualization: they do not emulate hardware, have no independent kernel, and directly share the host’s operating system kernel. They rely on Linux Namespaces for isolation and Cgroups for resource limiting, and are essentially a set of isolated processes on the host. Containers are therefore extremely lightweight — images are MB-sized, startup takes seconds, and deployment density is high. The downside is that the level of isolation is weaker than that of virtual machines: containers can only use systems with the same kernel as the host, and if the host kernel has problems, all containers are affected.

Name the switch models you have worked with

Huawei

1. Gigabit access S5735
2. 10G aggregation S6730
3. Chassis data center core CE12700

Lenovo

1. Gigabit server access G7052
2. 10G TOR leaf switch G8272
3. Chassis FC storage fiber switch B6505

H3C

1. Gigabit campus access S5560
2. 10G aggregation TOR S6520X
3. Chassis enterprise core S12500

Ruijie

1. Gigabit PoE campus access RG‑S5310
2. 10G aggregation RG‑S6150
3. Chassis data center core N18010

Brocade

1. Fiber SAN for small/medium G620
2. Fiber SAN for large G630

Cisco

1. Gigabit campus access C9200
2. 10G IDC leaf switch N9K‑C9300

Walk me through the entire rack-mounting process for a server

After the ticket workflow is completed, go to the data center and first inspect the server’s appearance and accessories to verify there is no damage and confirm it is normal Verify that the server model matches the rack position, prepare the rails and tools, secure the server in place with the power off, then connect the PDU power and rack cabling; once powered on, complete the initial configuration through the management port.