Elastic Compute Service
Elastic Compute Service
Document
- Documentation
Elastic Compute ServiceElastic Compute ServiceFAQInstance FAQ
Copy page
Download PDFFAQ
Instance FAQ
Copy pageDownload PDFInstance FAQ
FAQ on instance purchase
How can I view the regions and availability zones where instances are available for sale?
You can refer to Viewing regions where instances are available for sale to view the regions or availability zones where instances can be purchased.
What should I do when "Sold out" appears when I purchase an instance of a particular specification?
If the instance that you need to create is sold out, we recommend you adjust requirements based on your needs:
- Change the region or availability zone.
- Change the resource configuration and select an instance type with similar computing power and performance in the selected region.
Can I change the region and availability zone after purchasing an instance?
We recommend you select a region closest to your desired service when you buy an ECS instance. Note that once you have paid for an instance, you cannot change the region and availability zone.
How to select a region?
For good download performance and low latency, we recommend that you select a region closest to your desired service when you buy an ECS instance. Note that once you have paid for an instance, you cannot change the region.
How to select an availability zone?
- If your services require disaster recovery, we recommend that you deploy resources in different availability zones of a region.
- If low latency between your services is required, we recommend that you deploy resources in the same availability zone.
How do I choose an ECS instance suitable for my business?
How do I select the instance type when buying an ECS instance?
Is there a limit on the number of ECS instances that can be created at one time?
A maximum of 100 ECS instances can be created at one time. Currently, you cannot increase the instance creation quota.
How long does it take to use an ECS instance after I purchase one?
After creating an ECS instance, you can log in to and use it when the instance status is Running.
Can I buy ECS instances if my account does not undergo real-name authentication?
Why cannot I purchase Pay-as-you-go ECS instances?
If you want to buy a pay-as-you-go instance, the total value of your account balance (cash balance) and vouchers must be at least CNY 100.00.
How to purchase HPC-optimized instances?
You can purchase HPC-optimized instances in the ECS console. For detailed steps, refer to Purchasing HPC-optimized instances.
tip
HPC-optimized GPU instances are currently only available to authenticated enterprise accounts. If you need to create HPC-optimized GPU instances, undergo enterprise real-name authentication first. For steps to upgrade the personal authentication to enterprise authentication, refer to How can personal authentication be upgraded to enterprise authentication.
After creating an instance, I cannot find the newly created instance on the Instances page. Why?
This may be because the current project is different from the project associated with the instance. You can check the newly created instance from "All Resources" in the top navigation bar or from the project associated with the instance.
Why do I fail to select an existing subnet when creating an instance?
Possible causes are as follows:
- The availability zone associated with the subnet differs from the one where the instance is located.
- The subnet and instance are associated with different projects.
Why do I fail to select an existing security group when creating an instance?
Possible causes are as follows:
- The security group and instance are associated with different VPCs.
- The security group and instance are associated with different projects.
Is the EIP of the purchased ECS instance the same as EIP of the local PC?
No, the Elastic IP (EIP) of the ECS is provided by the cloud platform for the server, while the internet IP of the local PC is provided by the broadband line operator for your PC. These two IPs are independent and unassociated.
How to purchase multiple ECS instances?
You can purchase multiple ECS instances with the same cloud platform account. For operation details, see Purchasing and using ECS instances.
How to view the maximum quota limit for instances?
Can I buy ECS instances deployed in overseas regions?
ECS instances in the Asia Pacific region are currently available for purchase, where the network connections are relatively fast. This meets the needs of international customers and makes ECS instances better suited for users with business in other countries or regions. For the cities where data centers are located, refer to Regions and availability zones.
tip
- If you need to buy ECS instances deployed in overseas regions, contact the account manager to apply.
- For ECS instances deployed in Asia Pacific (Johor) and their prices, refer to Price calculator. The price calculator provides reference prices. The actual costs are the fee results displayed when you order in the ECS console.
FAQ on instance configuration change
What instance types can be changed?
Which instance types do not support changing?
Can I downgrade specifications of instances with high specification?
Yes, you can downgrade the specification for pay-as-you-go or subscription instances. For detailed steps, please refer to Changing instance type.
Can I upgrade specifications of instances with low specification?
Yes, you can upgrade the specification for pay-as-you-go or subscription instances. For detailed steps, please refer to Changing instance type.
Can I change the specifications of pay-as-you-go instances?
Yes, you can. You can upgrade or downgrade the configurations of purchased Pay-as-you-go instances. For specific operations, refer to Changing instance type.
Can I change the specifications of subscription instances?
Yes, you can. You can upgrade or downgrade the configurations of purchased subscription instances. For specific actions, refer to Changing the instance type.
Is hot modification (hot upgrade) available?
No. You need to stop the instance before hot modification. Otherwise, instance type changes cannot be made.
FAQ on instance usage
How to share an ECS instance with other users?
After you associate the instance with an EIP, other users can log in to it via remote connection tools. You need login credentials (password or key pair) and the instance EIP to do so. To select an appropriate login method, you can refer to Login method overview for comprehensive consideration, such as operating systems of the target instance and local host, and remote login tools.
Can I delete subscription instances?
When a Subscription instance expires or you unsubscribe from it, you can manually delete it within the resource retention period. If you need to release a subscription instance in advance, you can delete the instance after unsubscribing to it.
Can an instance be restored after it is deleted?
No, you cannot restore the deleted instance. Please proceed with caution. For details on the resource status after the instance is deleted, refer to Deleting instances.
How to view information of deleted instances?
System events are generated accordingly after an instance is deleted. To view system events on the Events page in the ECS console, refer to Querying and handling ECS events. Or, call the DescribeSystemEvents operation and then set Types.N to DeleteInstance.
How is the refund settled if I purchase an inappropriate instance?
If you purchase the inappropriate instance, the following solutions are for your reference:
- Pay-as-you-go instance: Billing is based on actual usage and cannot be refunded. You can delete the instance directly.
- Subscription instance: Currently, only non-7-day unconditional unsubscription is supported. For unsubscription steps, refer to Instructions on unsubscription. For unsubscription rules, refer to Instructions on non-seven-day refund policy.
If you purchased an instance at a discount, your eligibility for the discounted purchase cannot be restored after unsubscription. Please proceed with caution.
How to view the operating system version of an ECS instance?
You can refer to the following methods to view the operating system version of the instance:
- Method 1: ECS console
- Log in to ECS console, and then view the image version in the "Image" column on the Instances page or on the instance details page. For detailed steps, refer to Viewing instance information.
- Method 2: API operation
- Call the DescribeInstances operation. The returned OsName and OsType are the instance operating system name (such as CentOS 7.6 64-bit) and operating system type (such as Linux).
How to view the instance hostname?
You can view the hostname of the instance through the following methods:
- Method 1: On the Instances page in the ECS console, click the target instance name. Then you can view the instance hostname in the "Basic information" section on the Overview tab.
- Method 2: Log in to the instance, and then execute the hostnamectl command.
-
How to view the instance CPU model?
You can view the CPU model of the instance through the following methods:
- Method 1: On the Create Instances page in the ECS console, you can view the CPU model in the "Processor model" column of the instance configuration section.
- Method 2: Log in to the instance, and then execute the lscpu command.
-
How to view GPU graphics cards mounted to GPU instances and detailed numbers?
To view the types of graphics card mounted to instance families, refer to Instance types. To view the number of GPU graphics cards for different instance types, refer to the "GPU memory (GB)" column in the instance specification metric list.
How to view detailed information about a GPU graphics card?
- GPU instances on Linux operating systems: After logging in to the instance, you can execute the nvidia-smi command to query detailed information about GPU graphics cards.
- GPU instances on Windows operating systems: After logging in to the instance, you can view detailed information about the GPU graphics cards on the Device Manager.
How to install or uninstall GPU drivers?
GPU-based instances require installed drivers to enable the capabilities of GPU cards. Refer to NVIDIA driver installation guide to install plugins.
How to share a GPU graphics card?
You can share an mGPU through containers. This enables multiple containers to share one GPU graphics card and ensures performance and fault isolation. This supports flexible scheduling and strict isolation of computing power and memory, reduces GPU usage costs, and effectively ensures the efficient utilization of GPU hardware resources for training and inference.
Can I run any of my own programs on an ECS instance?
Like traditional hardware servers, ECS instances also provide basic infrastructure, such as operating systems, storage, network, and compute resources. After creating an ECS instance, you can install and configure software and applications that you need on the instance, like using your local PC or physical server. It should be noted that running programs on ECS instances requires compliance with usage regulations and security policies. Ensure your programs do not pose security threats to the ECS and other users. For details, refer to Usage notes.
Does an ECS instance provide a database by default?
ECS instances are not associated with databases by default. You can deploy your own database or separately purchase RDS for MySQL.
Can I install virtual machines on ECS instances?
No, you cannot install virtual machines on ECS instances.
Can I use external hardware devices on ECS instances?
No, you cannot attach external hardware devices to ECS instances directly.
Can I use sound card applications on ECS instances?
No, you cannot use sound card applications on ECS instances.
Why is an incorrect IP address displayed on a Windows instance after I mount, uninstall, or disable a network interface and perform other operations?
You can try to restart the instance. If the issue persists, execute ipconfig /release and ipconfig /renew commands within the instance to refresh the IP address.
What operating systems are supported for third-generation AMD instances?
For operating systems that are supported by AMD instances, please refer to Instructions on compatibility between AMD instance types and operating systems.
What should I note when I use Cloud Monitor, modify passwords, or use cloud assistant (Agent) for an instance?
When you upgrade plugins (Agent) provided by the platform, the systemctl daemon-reload command is executed automatically. To avoid business interruptions, please do not perform any operations on the instance that may cause conflicts.
Do ECS instances support secondary virtualization?
How to restore an ECS instance to its initial state?
To restore an ECS instance to its initial state, you can replace the operating system with the image that is used when you create the instance. For detailed steps, refer to Replacing operating systems.
tip
After you replace the operating system for instances, related data will be lost and cannot be recovered. We recommend you back up your data in advance with custom images or snapshots. For detailed steps, refer to Creating custom images based on instances or Creating snapshots.
Will the ECS instance shut down after I close the remote connection?
No, it will not. If you do not shut down the instance, the instance remains in the Running state.
Why does the Docker container in the ECS instance automatically run Ubuntu 18.04 image all the time?
The possible causes of this issue are as follows:
- Default image setting
- You may not specify the image to be used when creating the Docker container. Therefore, the container defaults to the Ubuntu 18.04 image from Docker Hub.
- Auto-start setting
- You may enable the auto-start setting when configuring your Docker container. Therefore, the container automatically starts and runs Ubuntu 18.04 image every time the instance starts.
- Orchestration tool setting
- If you are using orchestration tools such as Docker Compose or Kubernetes, these tools might default to the Ubuntu 18.04 image for container startup.
- Custom image error
- If you use a custom image but fail to build and deploy it correctly, Docker will then use the default Ubuntu 18.04 image.
You can try the following solutions to solve the problem:
- Check the startup command and configuration of the Docker container
- Make sure when creating or starting the container, you specify the required image rather than use the default Ubuntu 18.04 image.
- Set the automatic restart policy for the container
- If you enable the automatic start when configuring the Docker container, choose an appropriate automatic restart policy to handle unexpected situations.
- Check the configuration file of orchestration tools
- If you use orchestration tools such as Docker Compose or Kubernetes, make sure the correct image that you require is specified in the configuration file.
- Create and deploy the custom image correctly
- If you need to use a custom image, make sure the image is correctly built and pushed to the Container Registry.
Why is the website inaccessible via domain name but accessible via IP?
This may be because the domain name associated with your target instance is not filed. Please refer to Configuring domain name to complete operations.
Can ECS instances access overseas websites, such as Google, HuggingFace, and GitHub?
ECS instances do not restrict access to overseas websites, as long as the target website is accessible. Please note that when domestic servers in China are used for overseas website access, cross-border network issues may occur, such as packet loss, latency, or failure to access.
How to improve the read/write performance of ECS instances?
When read/write latency of ECS instances occurs, multiple factors are required to be considered, such as geographic location, instance configuration, cache technology, and load balancing. Detailed solutions are as follows:
- Select the optimal geographic location
- Select a region near your business to create an ECS instance to reduce data transmission time and latency. This can improve instance performance and accelerate access.
- Optimize read/write performance
- The read/write speed is associated with your ECS instance configuration. You can consider adjustments of instance compute resources, memory, and cloud volume configuration based on your needs. The platform provides a rich variety of instance types, which you can adjust based on your needs.
- Use cache technology
- Cache technology can be used to improve read/write latency. By caching frequently used data in memory, you can reduce the number of disk accesses and thus improve read/write speeds. The platform offers multiple cache services, such as Redis and MongoDB. You can select the appropriate cache technology based on your needs.
- Use Classic Load Balancer (CLB)
- If you have multiple ECS instances, you can consider CLB to distribute the load of read/write requests. CLB can distribute requests evenly across multiple ECS instances, thereby improving overall read/write speeds.
How to view the runtime duration of a program in an instance?
- Execute the following command to view the runtime duration of the target process.
Take viewing details of the runtime of "cloud-monitor-agent.service" (Cloud Monitor plugin) as an example. The start time is 2024-08-13 15:26:29 CST, and the current system time is 2024-08-13 15:38:18 CST, so the runtime of the cloud monitor plugin is 12 minutes.
- Click the "Windows" icon and then select "Windows System > Run".
- Enter the following content in the dialog box and click the "OK" button to open the Task Scheduler dialog box.
- Select "System Summary > Software Environment > Running Tasks" in the directory tree on the left.
- In the "Start Time" column on the right side of the target program, you can view its start time of execution. By comparing start time with the current time, you can obtain the program runtime duration.
Take viewing details of the runtime of "cloud-monitor-agent.exe" (Cloud Monitor plugin) as an example. The start time is 2024-08-12 05:45 PM, and the current system time is 2024-08-13 03:19 PM. Therefore, the runtime duration of the Cloud Monitor plugin is 1 day, 9 hours, 34 minutes.
What should I do if the instance lags?
If you find that the ECS instance is running slowly or suddenly experiences network disconnection, refer to the table below for troubleshooting and resolution.
How to deploy instances on different hosts?
You can use the deployment set feature and configure different deployment policies to meet this requirement. For details, refer to Deployment set overview.
What is the difference between the base frequency, Max Turbo Frequency, and all-core turbo frequency of an Intel processor?
- Processor Base Frequency: Also known as the default frequency, it is the reference frequency that the CPU can still achieve even under extreme conditions.
- Max Turbo Frequency: Also known as the single-core maximum frequency, it specifically refers to the maximum frequency that the highest-quality core within the CPU can achieve.
- All-core turbo frequency: Also known as the all-core maximum frequency, it refers to the maximum frequency that all physical or logical cores within the CPU can achieve at the same time.
Theoretically, the higher the Max Turbo Frequency, the stronger the single-core performance of the CPU is, and thus leading to higher efficiency in workload scenarios such as decompression. However, in a wider range of high-load application environments, all-core turbo frequency is actually the key metric that most significantly and directly affects CPU performance output.
tip
When your instance runs on a host machine with low load, you can get an all-core turbo frequency that is higher than the nominal full-core frequency. This can improve performance. However, all-core turbo frequency is not stable and changes with the load on the host machine. To keep the all-core turbo frequency of these instances at the nominal value, you can specify the CpuMaxFrequency parameter during instance creation.
Why is the total memory displayed by the free command or /proc/meminfo slightly less than the memory specified in the instance specification?
The total memory displayed by the free command or the /proc/meminfo is the available memory of the system.
Available memory refers to the total memory available for allocation to the kernel and applications after the operating system starts. Available memory excludes memory reserved by the BIOS for special purposes, as well as memory used for kernel code segments and metadata (struct page) for memory page management. Therefore, available memory is slightly less than the memory specified in the instance specification.
In addition, the memory used by kernel code segments and similar components varies slightly based on operating systems. Therefore, even when the memory specified in the instance configuration is the same, the available memory may differ slightly for instances with different operating systems.
How to migrate data from a Linux system volume?
When your business scales up, the existing system volume may no longer meet your needs. You can migrate system volume data to a data volume or another instance based on your needs.
- Migrate data from the system volume to a data volume.
- Execute the df -h command to view the usage rate of the system volume (for example: /mnt/eee).
- Mount the data volume for data storage to a directory (for example: /mnt/disk1) on an ECS instance.
- Copy the data from instance system volume (/mnt/eee) to the data volume (/mnt/disk1).
- Unmount the data volume from /mnt/disk1 and remount it to /mnt/eee.
- Execute the vi /etc/fstab command to modify the fstab file by adding the following content. This can specify the mount point of the data volume to enable auto-mount on boot.
- /dev/disk1 /mnt/eee ext4 defaults 0 0
- Restart the ECS instance to check if the instance runs normally. For example, verify if the database can be connected and accessed. If no exception occurs, the data migration is completed.
- Migrate the system volume data to other instances.
- Create a custom image for source instance to back up the instance data from system volume. For specific steps, refer to Creating custom images.
- Use the custom image of the source instance to create the target instance. After this, the system volume data can be migrated. For specific actions, refer to Purchasing an instance based on a custom image.
- Check the data storage and related data status within the newly created ECS instance. Make sure your business can operate normally after instance data migration. If no exception occurs, the data migration is completed.
How to upload local data to ECS instances?
If you need to upload local data to the ECS instance, please refer to File transfer overview, and choose an appropriate file transfer method based on the operating system type and your needs.
How to download data on ECS instances to a local computer?
If you need to download files on an ECS instance to your local computer, please refer to File transfer overview, and choose an appropriate file transfer method based on the operating system type and your needs.
How to migrate data from servers provided by other cloud vendors to VolcanoEngine?
You can use the Server Migration Center (SMC) to migrate the data on your own local servers, virtual machines, cloud hosts of other cloud platforms or other types of servers to ECS instances or ECS images (custom images) provided by VolcanoEngine. Learn more
Spot instance FAQ
What resources are covered by the price discount for spot instances?
Discounts only cover instance compute resources (vCPU, GPU, and memory). Prices for other resources such as volumes and Internet bandwidth are not discounted and depend on the prices of pay-as-you-go instances.
In what circumstances will a spot instance be released?
When resource inventory is insufficient or the spot market price exceeds the bid price, a created spot instance is automatically reclaimed by the system. For more details, refer to Reclaiming spot instances.
Is there a notification for spot instance release?
Yes. When a spot instance needs to be released due to supply-demand adjustments or price, the system sends a notification about releasing ECS instance resources before the instance is released (1 hour in advance for GPU instances and 5 minutes in advance for non-GPU instances). Notifications are sent via internal messages, SMS, and emails. When the time is reached, the instance is automatically released. At the same time, the system generates spot instance interruption events. To view the event details, refer to Viewing interruption events for spot instances.
Can I change the instance type of spot instances?
No, you cannot.
Can I change the billing method of spot instances?
No. The billing method of spot instances cannot be changed.
Why does the purchase of a spot instance fail?
- Instance resource inventory is insufficient.
- Your specified maximum price (the bid) is lower than the market price.
- The total value of your account balance (cash balance) and vouchers is less than CNY 100.00.
RI FAQ
What are RIs?
A voucher with deduction and reservation capabilities. Reserved Instances (RI) can be used to deduct the bills required by pay-as-you-go instances (excluding spot instances). You can also use RIs to reserve resources to ensure a stable supply. For more information, refer to Reserved Instances overview
What scenarios are applicable for RI?
- Automated O&M scenarios, where the number of instances requires automatic adjustments and the instance lifecycle needs flexible management based on business.
- The total resource amount remains relatively stable, but resources must be used flexibly based on business needs.
- If you need resource reservation, we recommend you purchase zonal RIs.
What is the billing method of RI?
You are required to pay for RIs separately. This means when creating RIs, you pay all fees as a one-off payment and no other fees are generated from RIs during the validity period.
Can I use RIs to deduct bills for spot instances?
No, you cannot change instance type for ESIs. You can use RIs to deduct bills for pay-as-you-go instances only.
Can I use RIs to deduct the storage and network parts of Pay-as-you-go instance bills?
No, you cannot change instance type for ESIs. You can use RIs to deduct bills generated by compute resources (vCPU, GPU, and memory) of Pay-as-you-go instances.
Can I reserve instance resources with RIs?
If you purchase zonal RIs, the system reserves instance resources for you with the same type and instance quantity as specified in the RIs. You cannot reserve resources with purchased regional RIs.
What instance types support the RI feature?
Can I change the instance type family specified in RIs?
No, you cannot change the specified instance type family of RIs. However, you can change instance type and quantity specified in RIs. For more details, refer to Modifying RI.
Can I use RI flexibly for different instance types across different availability zones?
Yes, you can use regional RI to deduct all Pay-as-you-go instances in the same region that belong to the same instance type family. Zonal RI can only deduct pay-as-you-go instances in the same availability zone that have the same specification and size.
Can I change a zonal RI to a regional RI?
Can I specify the pay-as-you-go instance to be deducted?
No, you cannot change instance type for ESIs. After you purchase RIs, they automatically match suitable pay-as-you-go instances within the validity period. You can not match them manually.
How to calculate the effective time of an RI?
After you purchase an RI, it takes effect immediately. The effective time is calculated on the hour. The billing starts from the effective time and ends at 23:59:59 of the expiration date.
For example, you have purchased an RI at 2022-11-01 22:45:00, with a validity period of one year. Therefore, the effective time and billing start time of this RI are 2022-11-01 22:00:00, and the end time is 2023-11-01 23:59:59. If you have a matching instance when you purchase an RI, the RI will be used to deduct the hourly bill from 2022-11-01 22:00 to 23:00 until the RI expires.
How to calculate the effective time of a modified RI?
After you modify an RI, a target RI is generated along with the immediate invalidity of the modified RI (original RI). The effective time of the target RI is calculated on the hour.
Suppose at 13:35:00 on December 20, 2022, you successfully changed an RI with the specification ecs.g2i.2xlarge (RI1) into two RIs with the specification ecs.g2i.xlarge (RI2 and RI3).
Then:
RI1 would expire at 13:35:00 on December 20, 2022. It could only be used to offset hourly bills for ecs.g2i.2xlarge instances generated before 13:00:00 on December 20, 2022.
RI2 and RI3 were also created at 13:35:00 on December 20, 2022. They could be used to offset hourly bills for ecs.g2i.xlarge instances from 13:00:00 on December 20, 2022 until they expired. Additionally, the resources they could offset and reserve would change from ecs.g2i.2xlarge to ecs.g2i.xlarge. If RI2 and RI3 successfully matched with an instance immediately after they took effect, they would start to deduct the hourly bill of the ecs.g2i.xlarge instance from December 20, 2022 13:00:00.
How to view RI deduction details?
You can refer to Viewing RI usage details to go to the billing center to check matching and deduction details between RIs and pay-as-you-go instances.
Can I get a refund for an RI?
Currently, RI only supports regular unsubscription and does not support unsubscription in seven days without reasons. For more information, refer to Renewal and unsubscription.
Can I associate RIs with tags or projects?
Yes, you can use tags or projects to manage RIs. For more information, please refer to Purchasing Reserved Instances (RI).
Bare Metal instance FAQ
Do bare metal instances support ECS instance creation?
Yes, they do. EBM servers offer complete processor capabilities at the physical machine level. However, creating VMs within a bare metal instance is not recommended, since this significantly degrades the performance of the created VMs. Instead, we recommend creating an ECS instance directly in the console.
Can I install virtual machines on EBM instances?
Yes. However, creating VMs within a bare metal instance is not recommended, since this significantly degrades the performance of the created VMs. For less impact due to performance loss, you can create a larger bare metal instance based on needs.
After creating a bare metal instance, why do I fail to log in to the instance via VNC?
During the startup of a bare metal instance, after the instance status changes to "Running", the operating system takes about one minute to start. At this stage, a VNC login may cause an error prompt.
After a bare metal instance is created, why does the uninstallation of volumes or network interfaces fail? And why do volumes or network interfaces needed to be uninstalled remain in the "Detaching" state?
During the startup of a bare metal instance, after the instance status changes to "Running", the operating system takes about one minute to start. At this stage, if you uninstall volumes or network interfaces, an error may occur. We recommend that you wait a few minutes before uninstalling volumes or network interfaces.
If you start to uninstall volumes or network interfaces immediately after the instance status changes to "Running", we recommend you stop the instance first. After this, the state of volumes or network interfaces can change to "Attached".
Shared instance FAQ
What does "Shared" in Shared instances mean?
The "Shared" in Shared instances refers to the CPU resource sharing mode rather than the sharing of IP address.
What is the difference between Shared instance types and Dedicated instance types (such as General-purpose, EBM, and GPU-accelerated)?
Based on whether CPU resources are shared, instances provided by the platform can be classified into shared and dedicated types. The two types have identical configuration options for storage (such as memory and system/data volumes) and network (such as Internet bandwidth), and differ only in CPU computing performance. Details are as follows:
Can I have ensured performance of Shared instances?
Shared instances have no computing performance stability ensured. Shared instances use an unassociated CPU scheduling mode where each vCPU is randomly allocated to a free CPU hyper-thread. The vCPUs of different instances compete for physical CPU resources, thereby causing unstable computing performance under high load. Shared instances mainly share the performance of computing resources at lower costs. However, computing stability is not ensured.
What impact does a DDoS attack have on Shared instances? How to respond to the DDoS attack?
A DDoS attack may cause the following impacts on Shared instances:
- Network performance degradation: A DDoS attack occupies network bandwidth with large amounts of malicious traffic. This causes blockage of normal business traffic, increased network latency, slower network speed, and even failure to establish new network connections, thereby affecting service response speed.
- Resource contention: In the context of Shared instances, a DDoS attack may cause the network bandwidth and computing resources of the instance to be occupied by malicious traffic. This may lead to tight CPU, memory, and other resources, thereby affecting the performance of other applications that run on the instance.
To mitigate the impact against DDoS attacks, we recommend the following measures:
- Enable DDoS Protection_Basic: By default, Cloud DDoS Protection_Basic protects all EIPs owned by users on the cloud.
- Enable DDoS Native Protection (Enterprise Edition): For large-scale DDoS attacks, DDoS Native Protection (Enterprise Edition) can effectively mitigate the impact on businesses and ensure normal service operation.
- Security system setup: You can also enhance the security of your cloud server by actions such as enabling HIDS security enhancement, configuring security groups, and increasing password complexity. For further information, refer to Enhancing ECS instance security.
- Optimize network architecture: You can use services such as Classic Load Balancer (CLB) and NAT gateways to improve system availability and security. This can reduce the risk of exposure to the public network at the same time.
Last updated: 2026.10.14 17:34:03
All rights reserved © Beijing Volcano Engine Technology Co., Ltd2024
