Ease of Use
Open-source compatibility: 100% compatible with open source Redis and Memcached protocols. You do not need to modify any code and can directly use open source clients to communicate with Tencent Cloud Distributed Cache database instances.
Installation-free use: You can directly request a Tencent Cloud Distributed Cache database instance on the Tencent Cloud platform purchase page, select the desired system architecture, with no need to install the instance manually, enabling convenient and fast use.
Cloud migration: Tencent Cloud distributed cache database provides comprehensive support for data migration in self-built scenarios, including Tencent Cloud self-built environments, VPN self-built, dedicated line self-built, IDC self-built, and other scenarios, supporting both cold migration and hot migration capabilities.
Ultra-High Performance
High throughput: The Standard Edition has a performance of up to 100,000+ QPS for concurrent response, and the Cluster Edition supports up to tens of millions of QPS as the number of shards increases. Their ultra-high performance can perfectly meet the needs in most business scenarios ranging from gaming, mobile apps to advertising and ecommerce.
Low latency: Tencent Cloud Distributed Cache delivers stable low-latency response, where 99% requests can be responded to within 1 ms or 10 ms under 50% or 90% load respectively.
Read expansion: In hot key scenarios, Tencent Cloud Distributed Cache allows dynamically increasing replicas to expand the read performance. It supports up to 5 replicas to read up to 500,000 hot keys.
Write expansion: The performance of the Cluster Edition equals to the shard quantity multiplied by the performance per shard and increases linearly as the shard quantity increases between 3 and 128.
Flexible Architecture
Tencent Cloud Distributed Cache supports multiple system architectures to meet the needs of various business scenarios. You can deploy the appropriate architecture based on your actual business requirements.
Standard architecture: Refers to the version that supports one or more replicas (where replicas denote non-master nodes), representing the most common distributed cache database version.
Cluster architecture: Adopting a distributed architecture, it supports scaling out and scaling down shards and replicas, offering high flexibility, availability, and performance of up to tens of millions of QPS.
Architecture upgrade: Supports upgrading from the standard architecture to the cluster architecture to meet the need of transitioning when the standard architecture faces insufficient performance and capacity.
Read/Write separation: Automatic read/write separation is implemented at the proxy layer. The proxy writes data only to the master node, while read requests are automatically routed to the configured read-only nodes by load balancing.
Multi-AZ deployment: It refers to the disaster recovery architecture where the master and replica nodes of an instance are deployed in different AZs in the same region to deliver a higher availability.
Global replication: Based on the original master-replica replication scheme, a new log file is added for remote replication to ensure the eventual data consistency for instances in different regions in the replication group. This solves the problem of cross-region data inconsistency and offers region-level disaster recovery capabilities.
Automatic disaster recovery: Tencent Cloud Distributed Cache adopts a hot standby architecture with dual servers. Upon host failure, access switches to the standby server within seconds without any user intervention. This saves the manpower and time costs required for developing master-replica systems.
Diverse Specifications
Tencent Cloud Distributed Cache offers capacity specifications of 0.25GB - 8TB, available in both standard and cluster editions.
Standard architecture: Supporting 1 - 9 replicas to meet varying requirements for availability and performance across different scenarios. When the number of replicas is ≥1, read/write separation can be enabled to improve read performance through replica nodes.
Cluster architecture: Supports horizontal scaling of 1-128 shards and 1-9 replicas. Scaling out, scaling in, and specification changes can be performed with minimal service disruption, ensuring maximum service availability.
Data Security
RDB persistence: The TencentDB backend service periodically creates snapshots for the data stored in the storage engine according to the backup policy to generate RDB files and then saves them in disks for data persistence.
Network security protection: Tencent Cloud Distributed Cache supports configuring security groups in VPCs to implement allowlist-enabled network access control, which ensures the security and reliability of network environments.
Account authentication: Tencent Cloud Distributed Cache allows you to authorize root accounts, sub-accounts, or across accounts, so you can manage your resources at a fine-grained level and benefit from enterprise-grade security protection.
Data isolation: Tencent Cloud Distributed Cache supports multiple layers of data isolation between different regions, AZs, networks, and account levels to ensure the data security and integrity.
Intelligent Ops
System monitoring: The instance usage process is fully transparent. It supports configuring alarm rules through Tencent Cloud Observability Platform (TCOP), providing over 30 automated monitoring metrics. This helps you maintain full visibility into the operational status of your cache database service and quickly troubleshoot issues.
TencentDB for DBbrain (DBbrain): It monitors and diagnoses database instance exceptions in real time (including slow logs, big keys, hot keys, and delays), automatically generates health reports, and gives expert optimization suggestions. This helps you promptly optimize the database performance.
Management and Control API Interfaces: Tencent Cloud Distributed Cache database provides a complete set of management and control API interfaces for implementing a series of resource self-management and Ops features.
Comparison with Self-Built Databases
Tencent Cloud Distributed Cache, provided as a service, offers significant advantages in flexibility and ease of use, high availability, fully managed Ops, and data security and reliability.
|
Performance | 80,000–100,000/shard. | 80,000–100,000/shard. |
Cost | You need to purchase hardware on your own, and the cost of a single storage server is relatively high. Building a high-availability master-slave architecture usually requires purchasing three devices, which can easily lead to resource redundancy. Additionally, you need to hire professional database administrators, resulting in high labor costs. | No hardware or software investment is required. Multiple instance specifications are provided, allowing you to request resources on demand with high resource utilization. Free Ops features including management, backup, cloning, monitoring, and alarms are offered, delivering higher overall cost-effectiveness. |
High Availability (HA) | Fault recovery time generally exceeds 60 seconds. | Failover time is less than 60 seconds. |
Availability | You need to build the primary-secondary replication architecture on your own and handle failures, resulting in high Ops complexity. | It supports multi-node hot standby, automatic disaster recovery and failover that are transparent to services, achieving 99.95% availability. It also supports prioritized reading from secondary nodes to ensure high-concurrency read performance. |
Scale-out/Scale-in. | Requires purchasing hardware, hosting data centers, and redeploying applications, resulting in a lengthy process. During data migration, brief outages and migration risks may occur. | One-click online scale-out/scale-in with rapid deployment, supporting no service interruption, no service suspension, and not read-only. |
Read-write separation | Requires implementation at the business code level. | Supports automatic read/write separation, which can be flexibly enabled or disabled based on business requirements. |
Resource isolation | Supports CPU and memory isolation. | Supports isolation across multiple dimensions such as CPU, memory, traffic, and number of connections. |
Multi-account | Not supported. | Supports multi-account management. |
Security Control | Not supported. | Supports a blocklist mechanism for disabling high-risk commands. |
Monitoring | Relies on Server collection with limited monitoring metrics. | Provides over 30 cluster and node monitoring metrics, including specialized monitoring for latency, traffic, large keys, and hot keys. Supports alarm reporting to facilitate autonomous Ops and risk prevention. |