How Leading Internet Companies Automate Operations: From Planning to Intelligent Management

1. Planning and Model‑Driven Design

To ensure smooth scaling, internet companies design their overall system architecture early with standardization and modeling in mind, treating new business resources like fast‑food orders that can be provisioned on demand.

Standardization: Use standard protocols and technologies for good extensibility and unified products, and adopt data‑center‑grade equipment for reliability, flexibility, and low latency.

Modeling: Design network architecture models based on business needs, validate them, establish baselines, and enable batch replication and unified management, which also facilitates automation.

2. Building Automation

After achieving batch replication capability, automation techniques boost deployment efficiency. Small teams (3‑5 people) can bring up new nodes; for example, a company dispatched two engineers to a remote site, then used central management to auto‑configure devices, completing installation within a week.

Key aspects are batch replication and automated onboarding.

Batch Replication: Identify technical concerns, design network models, test and pilot, produce hardware/software configuration templates for mass deployment.

Automated Onboarding: Leverage TR‑069, Autoconfig, and zero‑configuration features to onboard devices at scale.

Differences between Autoconfig and TR‑069:

Autoconfig is for zero‑configuration deployment and usually requires a dedicated network‑management system; TR‑069 provides a complete management solution, supporting ongoing monitoring, configuration, and software upgrades.

Autoconfig uses DHCP + TFTP (simple); TR‑069 uses DHCP + HTTP (complex) and needs a dedicated ACS server.

Security: TR‑069 is more secure, based on HTTPS/SSL.

H3C iMC BIMS implements the ACS function of the TR‑069 protocol, offering zero‑configuration capabilities, flexible networking, and management of DHCP and NAT‑behind devices. Its workflow is illustrated in Figure 3.

3. Intelligent Management

Network‑management teams need user‑friendly tools for information query and alarm handling. Early tools relied on command‑line interfaces and lacked batch support. Modern graphical and intelligent tools are preferred.

Graphical Management: Visual representation of data‑center hardware, network topology, server racks, and port connections.

Intelligent Management: Adopt new technologies to improve traditional MIB‑based management efficiency, introduce embedded automation architecture (EAA), and enable APP‑based management of smart terminals.

Current network‑management protocols include SNMP and Netconf. SNMP is simple and mature but falls short on security, efficiency, and complex operations. Netconf uses XML for configuration data, SSHv2 over TCP for transport, and RPC for operations, offering better reliability, security, and expressive modeling.

4. Platform‑Based Monitoring

By integrating basic monitoring tools such as Show, Display, SNMP, and Syslog, a unified monitoring platform can be built to achieve comprehensive visibility across the infrastructure.

5. Process‑Oriented Operations

Process‑driven operations help organize experience and improve issue‑resolution efficiency. Three key steps are:

Proactive Planning: Involve operations early in the planning phase to create a closed‑loop of mutual promotion.

Pre‑emptive Action: Leverage automation technologies like EAA to prepare automated response plans in advance.

Experience Learning: Build a historical knowledge base and apply lessons learned to avoid repeating mistakes.

Conclusion

Automation in operations is a major theme. With the rise of SDN, server virtualization, and other technologies, network and server management are undergoing significant change. Practices based on ITIL have yielded valuable experience, and more enterprises are innovating in automated operations. H3C’s management platform is evolving, and future platforms will likely integrate SDN, virtualization, and other advanced technologies to achieve truly efficient automated operations.