Preparations Before Buying a DCS Instance
Before purchasing a DCS instance, identify your requirements and complete the following preparations:
- Decide on the required cache engine.
Choose a cache engine based on service requirements. The cache engine cannot be changed once the instance is created.
- Decide on the required cache engine edition and version.
Perform this step if you choose Redis as the cache engine.
DCS provides the standard and high-performance editions of Redis. The high-performance edition uses Huawei proprietary lightweight LibOS. Different Redis versions have different features. For details, see Comparing DCS Redis Versions.NOTE:
The high-performance edition is supported only in CN North-Beijing1.
- Decide on the required instance type.
DCS provides single-node, master/standby, Proxy Cluster, and Redis Cluster types of instances. Each type has its own architecture. For details about the instance architectures, see DCS Instance Architecture.
- Decide on the required instance specification.
Each specification specifies the maximum available memory, number of connections, and bandwidth. For details, see DCS Instance Specifications.
- Decide on the region and whether cross-AZ deployment is required.
To minimize latency, you are advised to select a region closest to your application. For example, if your application is in Guangzhou, select CN South-Guangzhou.
A region consists of multiple availability zones (AZs) with physically isolated power supplies and networks. Master/Standby and cluster DCS instances can be deployed across AZs. Applications can also be deployed across AZs to achieve high availability (HA) for both data and applications.NOTE:
- If a master/standby or cluster DCS instance is deployed across AZs, faults in an AZ do not affect the cache node in other AZs. This is because when the master node is faulty, the standby cache node will automatically become the master node to provide services. Such deployment achieves better disaster recovery.
- Deploying a DCS instance across AZs slightly reduces network efficiency compared with deploying an instance within an AZ. Therefore, if a DCS instance is deployed across AZs, synchronization between master and standby cache nodes is slightly less efficient.
- Decide whether public access is required.
Currently, public access can be enabled only for DCS Redis 3.0 instances. You can configure whether to access the instance over public networks through SSL. For details, see Can DCS Be Accessed over Public Networks?
- Check whether EIPs are available.
Public access requires EIPs. If you decided that public access is not required, you do not need to purchase EIPs.
- Decide whether backup policies are required.
Currently, backup policies can be configured only for master/standby and cluster DCS instances. For details about backup and restoration, see Overview.