AI Assistant
Contact UsDocsConsole

Product Feature

Elastic Compute Service is a secure, reliable, and scalable cloud computing service that helps you reduce IT costs and improve O&M efficiency.

Buy NowFree Trial

Elastic Compute Service (ECS)

    product feature

    Global infrastructure

    • Multiple regions and multiple availability zones
      Alibaba Cloud’s infrastructure currently spans four major continents, with more than 20 public cloud regions and over 80 availability zones in operation worldwide. In addition, Alibaba Cloud offers dedicated regions for financial and government clouds, and is committed to continuously planning and building new regions to better meet the diverse business and scenario requirements of its users. With the accelerated investment in and deepening deployment of its infrastructure, Alibaba Cloud is laying a solid foundation for users to enjoy a high-quality cloud computing experience. Deploying instances within different availability zones in the same region provides higher disaster recovery capabilities, while creating instances within the same availability zone results in lower network latency between instances, which can improve user access speeds. You can choose regions and availability zones based on factors such as the geographic location of your users, the availability of Alibaba Cloud products in specific regions, the required application availability, and whether internal network communication is needed, to meet your business needs. For more information, see Regions and Availability Zones.
    • Global Time Synchronization Service
      Alibaba Cloud’s NTP server supports time protocols such as NTP and PTP, meeting a wide range of time synchronization needs and providing nanosecond-level precise time synchronization services for Elastic Compute Service (ECS) instances. The server also boasts excellent fault tolerance: multiple Alibaba Cloud NTP servers can be grouped into a time synchronization cluster to achieve fault tolerance and load balancing. For more information, see the Time Synchronization Service.
    • A variety of computing architectures and instance types
      Alibaba Cloud servers support a variety of architectures, including x86 and Arm computing, GPU-accelerated heterogeneous computing, Elastic Bare Metal instances, and Super Computing Clusters. They offer hundreds of instance type families to meet the needs of users of different scales and workloads, allowing you to choose the most suitable cloud server based on your specific use case. For more information, see Instance Type Families.

    Cost optimization

    • Multiple billing methods
      Elastic Compute Service (ECS) offers a variety of billing methods, including pay-as-you-go, savings plans, and preemptible instances. Pay-as-you-go allows you to provision and release resources on demand, eliminating the need to purchase large amounts of resources in advance. This can reduce costs by 30% to 80% compared to running a self-managed data center.

      Savings plans are a pay-as-you-go discount program designed for long-term, stable resource usage. By committing to a consistent level of consumption over an extended period, you can enjoy discounts of up to 70% off the standard pay-as-you-go rate.

      Preemptible instances offer significantly lower prices for non-core workloads and are ideal for time-consuming computing tasks that can be interrupted without causing major issues. In addition, you can combine pay-as-you-go with subscription-based plans and reserved instance coupons to further optimize your resource usage costs.

    • Auto Scaling and Elastic Provisioning
      With Auto Scaling, you can automatically adjust the number of servers based on your business needs and workload. When demand increases, Auto Scaling automatically adds instances of a specified type to ensure sufficient computing power. Conversely, when demand decreases, Auto Scaling automatically reduces the number of instances of the specified type to help reduce costs. Leveraging its elastic provisioning capabilities, the service can automatically create preemptible instances of various instance types or combine pay-as-you-go and preemptible instances, enabling you to deliver stable total computing power at the lowest possible cost.
    • Stop-Start Mode
      If you need to shut down your instance for an extended period, you can enable Economical Mode. Once enabled, you will no longer be charged for compute resources (vCPUs and memory) or the fixed public IP address. While retaining the data and configuration information of your pay-as-you-go instance, this mode helps you save on resource usage costs.

    Network

    • Performance optimization
      Built on Alibaba Dragonwell, with deeply optimized JVM parameters, resulting in significant improvements in performance metrics such as TPS and QPS.
    • Build highly available, multi-network cloud services using Elastic Network Interfaces (ENIs).
      An ENI is a highly flexible virtual network interface that provides ECS instances with network interfaces and IP addresses, which can be freely bound to or unbound from the instance. You can attach multiple ENIs to an ECS instance to achieve multiple IP addresses, multiple network interfaces, high network availability, traffic isolation, and other advanced networking capabilities.
    • Use Elastic IP Addresses to dynamically manage IP addresses.

      Elastic IP Address (EIP) is a dynamically allocated public IP address that can exist independently of an ECS instance. It can be unbound from an ECS instance at any time and re-bound when needed, making it ideal for scenarios where public network communication needs to change frequently while the IP address remains constant.

    • Use PrivateLink to establish secure and stable private connections with services on Alibaba Cloud.
      PrivateLink enables you to establish secure and stable private connections between your Virtual Private Cloud (VPC) and services hosted on Alibaba Cloud, simplifying your network architecture and allowing you to access services over a private network while avoiding the potential security risks associated with accessing services over the public internet.
    • Security groups control inbound and outbound traffic and define security domains.
      A security group is a virtual firewall that controls inbound and outbound traffic for ECS instances, enabling you to configure network access control for one or more cloud servers. Security groups provide stateful packet inspection and filtering capabilities, allowing you to define security domains in the cloud based on the features of security groups and the rules you configure within them.
    • Enhance network performance with features like jumbo frames and eRDMA.
      By increasing the frame size, jumbo frames enable each packet to carry more data, thereby reducing the number of packets that need to be transmitted. This, in turn, lightens the load on the processor and boosts overall data transfer speeds. eRDMA is an elastic RDMA network for the cloud developed in-house by Alibaba Cloud. Leveraging Alibaba Cloud’s X-Dragon architecture MoC card, eRDMA delivers ultra-low latency, high throughput, low CPU overhead, and highly scalable high-performance RDMA networking services. It supports large-scale RDMA networking within seconds and is compatible with both traditional HPC applications and conventional TCP/IP applications.

    Operations and maintenance

    • Performance optimization
      Built on Alibaba Dragonwell, with deeply optimized JVM parameters, resulting in significant improvements in performance metrics such as TPS and QPS.
    • Detect underlying ECS infrastructure anomalies in a timely manner through system events.

      System events are used to record and notify you about information related to cloud resources, such as whether a resource has encountered an anomaly or whether its status has changed. System events also provide O&M capabilities that enable rapid recovery of faulty instances. In addition, system events offer subscription capabilities, allowing you to build event-driven, automated O&M workflows.

    • Instance health diagnostics to quickly identify and resolve ECS access issues with a single click.
      The Instance Health Diagnosis feature performs a comprehensive diagnosis of the instance’s running status, network status, disk status, and relevant configurations within the instance’s operating system, enabling you to quickly identify the root cause of why the ECS instance is inaccessible.
    • Real-time monitoring of the Guest OS’s operating status through instance health status.
      The instance health status reflects whether the instance’s operating system is running normally and enables you to promptly detect issues such as OOM or blue screens on the instance.
    • Achieve high availability and low latency for ECS instance deployment through deployment sets.

      Deployment Sets are a service that helps you manage instance deployment policies. Deployment Sets support low-latency networking policies, enabling you to deploy ECS instances within a single network topology to minimize network latency between instances. They also support high-availability policies, which ensure that ECS instances are strictly distributed across different physical servers to achieve high availability.