Stack is a cloud native – site-based data central-platform PaaS, and we have an interesting open source project on Github and Gitee: FlinkX is a batch data synchronization tool based on Flink. It can collect static data and real-time data. It is a data synchronization engine with all-domain, heterogeneous and batch integration. If you like, please give us a star! Star! Star!

Github open Source project: github.com/DTStack/fli…

Gitee Open Source project: gitee.com/dtstack_dev…

— — — — — — — — — — — — — — — — — — — — —

More and more database products, cloud manufacturers have launched relational database products based on open source MySQL, Postgre, etc., multi-copy, high availability, read and write separation, library and table functions are integrated into all kinds of products, reducing the construction and operation and maintenance costs of computer room, helping more customers to access the cloud.

Oracle’s Cloud output is rare, except for the Oracle Cloud.

For the current business system based on Oracle system development, and difficult to transform, such as OA, ERP and other systems; Or the database requirements are high, must use Oracle database scenarios, such as order flow, financial delivery system, Kangaroo cloud can provide Ali, Huawei, Tencent public cloud, Ali, Zstack private cloud perfect Oracle RAC cloud implementation deployment, operation and maintenance scheme, to assist customers seamless cloud access.

There is no difference between Oracle deployment on a single cloud or in a cluster and the official guidance documents.

However, in the basic configuration, the cloud environment outputs network, server and disk as products, which has its particularity:

  • Cloud switch: It is recommended that all service systems in a cloud environment run in a private network VPC and create switches on different network segments for RAC clusters and service systems. The same VPC network can communicate with each other by default
  • Cloud server: In a cloud environment, different availability zones in the same region are similar to different equipment rooms in the same city. Cluster computing nodes must reside in the same availability zone. You can purchase machines in different availability zones in the same region to deploy service/database Dr Environments
  • Cloud disks: Purchase and mount a cloud disk in the same region and available area as the RAC compute node based on the plan.

Different cloud environments have different limits on the number of disks mounted to a cloud server. Therefore, you need to consider the estimated data volume and cloud disk size during cluster planning.

Cloud server specifications can be configured one by one to reduce the impact on services.

Overview of RAC planning on cloud

Two, stable and efficient operation

1. Standard stress test

Swingbench can build real OLTP transaction scenarios, such as business, new customer order class, order, order process operation, through the mass increase and delete operation, testing different concurrent scenarios trend of database performance and stability, combined with other monitoring tools at the same time, to observe the performance of server memory, load and disk trend and stability; Based on the results of pressure measurement, early detection and optimization of problems, evaluation of the best use of the database.

According to the kangaroo cloud cluster operation and maintenance experience, a short time and low concurrency pressure test cannot verify the stability of the cluster. Therefore, after cluster deployment on the cloud, a long time (more than 4 hours) and high concurrency continuous pressure test will be carried out.

At the same time, it analyzes the cluster performance problems measured by pressure, optimizes the configuration of cluster parameters, and proactively provides suggestions on cloud server and storage resources optimization, in order to deliver a stable and high-performance database environment.

Aliccloud 2 nodes 12.2.0.1RAC pressure test example:

2. Service playback test

DBReplay is a test method that captures service loads during a certain period of time in the production environment. A single capture allows multiple replays, which is more intuitive and effective. Mainly used for database upgrade, migration, application deployment, etc., to quantify the impact of the implementation of the change on the existing performance.

For normally stressed business systems, such testing is necessary before migration cutover.

Iii. Case sharing

1. Some day enterprise customers

(1) Demand analysis

  • The overall service is in the cloud, and one Oracle RAC 12C or 11G must be deployed on the cloud
  • Meet requirements such as high availability and disaster recovery

(2) Architecture planning

  • Planning of 20 g3 OCR/Voting group for cloud disk, 60G2 Cloud disk redodg1/2 group, 200G1 Create an ARCHDG group for cloud disks and plan 500 GB1 Cloud disk As the DATADG group, 100 GB x 1 cloud disk as the MGMTDG group
  • Plan cloud service 32C64G x 2 as the compute node
  • You are advised to use the service separation mode to access the database

2. A recruitment website of a state-owned enterprise

(1) Demand analysis

  • Need to meet 1W people online access at the same time, more than 2000 concurrent transactions
  • Build Oracle standalone database on cloud by myself, which is expected to reach the bottleneck soon
  • A high availability and Dr Architecture is required to ensure service continuity

(2) Architecture planning

  • Newly purchased high-specification cloud ECS as computing nodes to ensure sufficient computing resources
  • The original stand-alone database is cutted and run as the DG standby database
  • Configure a proper backup policy and upload the backup set to the OSS object storage for remote storage

3. A provincial proprietary cloud traffic management platform

Collect provincial traffic management data, and provide a platform for inquiry and analysis.

(1) Demand analysis

  • A self-built equipment room is deployed on a single Windows server Oracle11.2.0.1
  • 18T data, and gradually rising
  • Business debugging requires docking and adjustment with the Ministry, so there is no possibility of testing
  • Short maintenance time

(2) Architecture planning

  • Deploy in a private cloud environment, two-node RAC cluster, and set up a Dr Database
  • Through ADG real-time synchronization machine room data, DBRelay pressure measurement
  • Use the fusioncube backup machine to back up production data