Huawei Cloud Database Services Overview
Huawei Cloud Database Services Overview
Foreword
⚫ After learning basic compute, storage, and networking services on Huawei
Cloud, let's move on to database services.
RDS DDS
EVS OBS
Traditional applications require databases that New Internet applications require databases
Application model can process transactions, complex queries, and that can store massive data and handle a large
joint queries, and maintain data integrity. number of concurrent requests.
10 ms-level
ms-level
Access performance Depending on query complexity and the
Irrelevant to the amount of stored data
amount of stored data
Semi-structured (key-value pairs and
Data types Fully structured
documents)
Vertical scaling
Vertical scaling
Scalability Horizontal scaling: primary instance + read
Horizontal scaling: cluster + read replicas
replicas
Database ecosystem
Building complete database solutions
AZ AZ1 AZ2
Failover Failover
Backups in OBS
Latest automated
backup
Latest Binlog 01 03
1 2 3 Log backups
OBS
Write Read
Basic functions
request request
Read/write splitting address ⚫ A unified read/write splitting address is provided,
(proxy)
transparent to applications.
⚫ Read weights are configurable for each node.
Write ⚫ If a health check detects that a read replica fails or
request
Read Read Read
request request request the I/O latency of a read replica goes beyond your
HA
preset threshold, read requests are no longer
distributed to the read replica.
deleted as needed.
RDS
Standby ⚫ The number of read replicas
App Server
Read replica supported per DB instance varies
RDS
Primary
Application
RDS console
DBAs Read-only
Read/write
Instance management
Instance access Backups
Backup and restoration
Instance monitoring Monitor Primary Failover Standby Read replicas
Scaling instance instance
Log management
Replication
Parameter
configuration
Delivery and O&M
personnel
Storage Storage Storage
⚫ Huawei DB engine
⚫ Compatible with MySQL clients
⚫ Decoupled storage and compute
⚫ Cross-AZ HA
GaussDB(for MySQL)
⚫ Three copies stored to ensure reliability
⚫ Scale-out and scale-up
Read replica Primary node Read replica ⚫ Compute and storage are decoupled,
Compute improving performance to several times that
layer
of open-source MySQL.
DB DB
DB DB DB DB
DB DB DB
GeminiDB
⚫ High availability
Application Application Application Application Application
If a compute node fails, workloads can
be switched over within seconds.
Three copies of data are stored, so the
Cassandra API Mongo API Influx API Redis API HBase API system can run fine if a storage node
fails.
⚫ Unlimited scalability
Compute nodes are stateless and can be
Node 1 Node 2 Node 3 Node 4 . . . Node 5 quickly scaled out.
Shared storage means you can add
nodes in a click or two and without
interrupting services.
01 03
There are no
logical
Users can locate
Each data relationships
data using key
record is a between the
values rather than
structured data records in a
through complex
document. collection.
queries.
02
⚫ O&M functions
⚫ No client required Developers DBAs ⚫ Host and instance performance
⚫ Visual operations data analysis
⚫ Online access ⚫ Slow and full SQL statement
analysis
⚫ Databases and table
creation, SQL statement ⚫ Real-time database performance
execution. diagnosis and analysis
⚫ Historical data analysis
Huawei Cloud
Developer Institute