HCIE-Cloud flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

ECS flavor naming

A flavor such as c6.large.2 encodes family (c = compute-optimized), generation (6), size class (large), and vCPU count (2). Memory is determined by the family ratio, not the name suffix.

EVS disk type selection

Common I/O suits low-load test systems, high I/O suits general business, ultra-high I/O suits high-throughput databases, and extreme SSD suits latency-sensitive workloads. Baseline IOPS rises with capacity for the same type.

VPC scope

A VPC is created within a single region and contains subnets that may span different AZs. Instances in different VPCs cannot communicate by default and need a VPC peering connection or VPN/Cloud Connect.

Security group vs network ACL

Security groups are stateful and applied at the instance or ENI level with allow-only rules. Network ACLs are stateless and applied at the subnet level with both allow and deny rules.

Dedicated vs shared ELB

Dedicated load balancers provide isolated resources, higher performance, and multi-AZ reliability. Shared load balancers are cost-efficient basic layer 4/7 load balancing with fewer features and lower guaranteed performance.

ELB listener and backend group

A listener binds protocol and port on the load balancer front end. A backend server group holds the ECS instances and health check configuration that receive forwarded traffic.

OBS storage classes

Standard for frequent access, Infrequent Access for less frequent reads, Archive and Deep Archive for long-term retention. Lifecycle rules automatically transition objects between classes based on object age.

Auto Scaling group bounds

A scaling group defines minimum, maximum, and expected instance counts. Scaling policies (alarm, scheduled, periodic) adjust the count, and the group never exceeds the configured bounds.

OBS vs EVS vs SFS

OBS provides object storage over HTTP APIs for unstructured data. EVS provides block storage attached to one ECS. SFS provides shared file storage mountable by multiple ECS instances simultaneously.

IMS private image

A private image is created from an existing ECS or an uploaded image file and can be used to launch new ECS instances with identical software configuration, enabling rapid horizontal scaling.

ECS vertical scaling

Changing an ECS flavor to a larger or smaller specification. It requires stopping the instance first, so it is not suitable for continuous-availability workloads; use Auto Scaling for horizontal scaling instead.

BMS bare metal server

A physical server offered as a cloud service with no hypervisor layer, giving full hardware performance and isolation. It supports VPC and ELB integration like ECS but cannot be live-migrated.

VPC peering connection

A connection between two VPCs, in the same or different accounts and regions, that allows private IP communication. Route entries must be added in both VPCs for traffic to flow.

NAT gateway function

Enables ECS instances in a private subnet to access the internet using an EIP without exposing their private IPs. It supports SNAT for outbound and DNAT for inbound port forwarding.

EIP elastic IP

A public IP address that can be bound to or unbound from ECS, ELB, NAT gateways, and BMS. Bandwidth can be billed by bandwidth or by traffic, and it can be shared via a bandwidth pool.

OBS bucket policy

A JSON policy attached to a bucket that grants or denies actions for specified principals, including anonymous users. It works together with IAM policies, and explicit deny takes precedence.

IAM agency

A delegated identity that grants a cloud service or another account temporary permissions to access resources, avoiding the sharing of permanent access keys. Commonly used for ECS to access OBS.

IAM policy evaluation order

Permissions are the union of all Allow statements from attached user and group policies, but any matching explicit Deny overrides all Allow statements, resulting in the request being rejected.

Region and AZ

A region is a geographic area containing multiple AZs. AZs within a region are physically isolated with independent power and network, so cross-AZ deployment protects against single-datacenter failure.

RPO and RTO

Recovery Point Objective is the maximum tolerable amount of data loss measured in time. Recovery Time Objective is the maximum tolerable service downtime after a disaster occurs.

CBR backup service

Cloud Backup and Recovery creates backups and vaults for ECS, EVS, SFS, and databases. It supports policy-based scheduled backups and restores entire instances or individual disks.

SDRS storage disaster recovery

Replicates ECS instances and EVS disks to a protection group in another AZ or region. On failover, protected instances start at the DR site with the replicated disks attached.

CES Cloud Eye alarms

A monitoring service that collects metrics for cloud resources and triggers alarm rules. Alarm actions can send notifications or invoke Auto Scaling to add or remove instances.

CTS audit trail

Cloud Trace Service records API calls made in the account, including the caller identity, source IP, time, and result. Traces are retained in a configured OBS bucket or LTS.

SMS server migration

Server Migration Service migrates physical servers, VMs, or other cloud hosts to Huawei Cloud ECS by installing an agent on the source and syncing disks incrementally before cutover.

DRS database migration

Data Replication Service migrates databases between on-premises, other clouds, and Huawei Cloud. It supports homogeneous and heterogeneous engines and can perform online incremental synchronization.

OMS object migration

Object Migration Service moves object data from other object storage platforms or HTTP sources into OBS, supporting migration task scheduling, traffic control, and consistency checks.

Active-active DR

Two or more sites serve traffic simultaneously through global load balancing. It gives near-zero RTO and RPO but requires application-level data synchronization and conflict handling.

Backup and restore DR

The simplest DR strategy: data is backed up to a secondary location and restored after a disaster. RPO depends on backup frequency and RTO is high due to restore time.

Pay-per-use vs subscription

Pay-per-use bills by actual usage time and suits short-term or unpredictable workloads. Yearly/monthly subscription offers lower unit price and suits stable long-running production workloads.

KooCLI command line

Huawei Cloud command-line tool that calls cloud APIs for scripting and automation. It uses AK/SK or temporary credentials and supports profile switching between regions and accounts.

Cloud Connect

A dedicated private connection between an on-premises data center and Huawei Cloud. It provides higher bandwidth and lower latency than internet VPN and is used for hybrid cloud architectures.