Alibaba Cloud postpaid billing account Alibaba Cloud Server Speed Test Guide: How to Measure ECS Download and Upload Speeds Before and After Purchase
Most users searching for this topic are not just looking for a command like speedtest or iperf3. They are usually trying to answer practical questions before spending money or before migrating workloads: Can my account buy Alibaba Cloud ECS immediately? Do I need identity verification? Which payment method works in my region? Why did my order go into manual review? Which ECS bandwidth setting should I choose if I want to test real upload and download performance? Why is the measured speed lower than the purchased bandwidth?
This guide is written for those decisions. It covers the full path from account preparation to speed testing and troubleshooting, with emphasis on purchasing, KYC, risk control, account restrictions, cost comparisons, renewals, and actual ECS test steps.
What users usually want to know before running an Alibaba Cloud speed test
In real operations, the questions behind this search usually look like this:
- Alibaba Cloud postpaid billing account Can I buy an ECS instance with public bandwidth using my current Alibaba Cloud account?
- Does Alibaba Cloud require identity verification before purchase?
- What payment methods are available, and can I prepay or pay-as-you-go?
- Will a new account, foreign payment card, or cross-border purchase trigger risk review?
- How do I create an ECS instance that can actually be used for upload and download testing?
- Which region should I select if my users are in Southeast Asia, Europe, the US, or mainland China?
- How do I test server-side throughput without confusing it with my local broadband speed?
- Why is the speed test showing far below the bandwidth I purchased?
- How much will bandwidth cost, and how do renewals affect my public IP and test setup?
If your goal is to make a purchase decision or validate whether Alibaba Cloud ECS fits your workload, those are the right questions to solve first.
Preparation: account, verification, funds, permissions, and resources
Account requirements
Before you can test Alibaba Cloud server download and upload speeds, you need an Alibaba Cloud account that is allowed to purchase ECS and related network resources. A newly created account may be able to log in but still be restricted from ordering certain billable products until verification or payment checks are complete.
| Item | What to prepare | Notes |
| Alibaba Cloud account | Registered email or phone | Use stable contact details because order verification notices may be sent there |
| Identity verification | Individual or business verification | Some orders may remain pending until verification passes |
| Payment method | Card, bank-supported payment channel, or account balance | Available methods vary by country and billing entity |
| Billing mode decision | Subscription or pay-as-you-go | For short tests, pay-as-you-go is usually simpler |
| Operational access | Console access with ECS creation permissions | In enterprise accounts, RAM sub-accounts may need explicit permissions |
Identity verification and KYC
For many users, the first practical blocker is not the speed test itself but identity verification. If the account has not completed the required real-name or business verification, ECS purchasing may fail or enter review. This is especially common when:
- The account is newly registered and attempts a high-value order immediately
- The payment card country does not align with the account profile
- The selected region or billing entity requires additional compliance review
- You are purchasing products that include public network resources
To reduce delays, complete verification before attempting the order. Prepare the exact legal name, identification details, and for business accounts the company registration information. Mismatched names between account identity and payment instrument can trigger review.
Payment methods, funding, and renewals
Users comparing costs should decide billing mode before creating the test instance.
| Billing option | Best for | Operational impact |
| Pay-as-you-go | Short-term speed tests, temporary migration validation | Easy to release after testing; watch hourly resource charges and bandwidth billing |
| Subscription | Longer evaluation or production workload | Lower unit price in many cases, but less flexible if you only need a short test |
| Account balance funding | Teams that prefer prepaid spending control | Helps avoid failed renewal due to card issues |
Renewals matter if you keep the ECS running for monitoring or repeated throughput tests. If a subscription instance expires, service interruption can affect test continuity. If your public IP is tied to the instance configuration, releasing and recreating resources can also change the endpoint used in your historical comparisons.
Risk control and compliance reviews
Alibaba Cloud may place orders under risk control review. Users usually encounter this when they are trying to buy quickly for urgent testing. Common triggers include:
- Multiple failed payment attempts
- High-value purchases on a new account
- Repeated creation and release of public-IP-enabled instances in a short period
- Mismatch between account region, IP login history, and payment origin
- Use of anonymous or inconsistent profile information
If your order is pending review, do not keep retrying with different cards repeatedly. That often increases the review probability. Instead, verify the account profile, ensure your contact email works, confirm cardholder and account identity consistency, and wait for the support or review request. For enterprise purchases, using a verified company account usually reduces friction.
How to buy an ECS instance suitable for bandwidth testing
If your goal is to measure Alibaba Cloud server upload and download speeds, the ECS configuration must be selected carefully. Many inaccurate tests come from choosing an instance without the right public network setup.
Which parameters matter most
| Parameter | Recommendation for testing | Why it matters |
| Region | Choose close to your target users | Bandwidth alone does not determine actual throughput; latency and routing matter |
| Availability Zone | Any zone in the chosen region unless integrating with existing VPC resources | For basic internet speed tests, region matters more than zone |
| Instance type | At least 2 vCPU / 2-4 GB RAM for clean testing | Very small instances can bottleneck encryption, compression, or traffic generation tools |
| System disk | 40 GB or larger SSD class disk | Disk speed should not interfere if you test with file transfer tools |
| Public bandwidth | Set explicit outbound/inbound internet capacity as available in the purchase page | Without public bandwidth, internet speed tests are not meaningful |
| OS | Ubuntu 22.04, Debian 12, CentOS-compatible distribution, or Windows Server | Linux is simpler for command-line throughput tools; Windows fits RDP users |
Cost comparison for a test-focused purchase
For pure speed measurement, avoid overbuying CPU and disk. Spend on the network setting that matches your real expected traffic. In many cases, users purchase a larger instance but keep a small public bandwidth cap, then misread the results as server weakness. The test result is usually limited by the bandwidth plan, route quality, remote endpoint, or local ISP path.
A practical approach:
- Short test: pay-as-you-go ECS with moderate compute and the specific public bandwidth you want to validate
- Migration rehearsal: same region and same bandwidth tier you plan for production
- Cross-border site test: create two or more instances in candidate regions and compare latency plus throughput
Step-by-step: create the instance and configure it for speed testing
Step 1: Create the ECS instance in the console
In the Alibaba Cloud console, open the ECS purchase page. Select the billing method first. If you are only evaluating network quality, start with pay-as-you-go to avoid committing to a long subscription before you confirm performance.
Where to click and what to configure:
- Open the ECS purchase page from the console product menu
- Select the target region based on where your users are located
- Choose an availability zone if required; for standalone testing any available zone is fine
- Select an instance family with at least 2 vCPU to avoid CPU bottlenecks during encryption-heavy transfers
- Choose Linux or Windows according to your testing method
- Set the system disk size and disk class
- Enable and configure public bandwidth or public IP-related internet access options shown during purchase
- Set login credentials: password or SSH key for Linux; password for Windows
Important considerations:
- If your account is under enterprise control, verify the RAM permissions allow ECS creation, VPC usage, and security group editing
- Record the selected region and bandwidth settings so you can compare later with measured throughput
- For cross-border workloads, do not assume the geographically closest region is always the fastest; routing quality differs by ISP
Step 2: Configure networking, public IP, security groups, and firewall rules
After the instance is created, confirm it actually has internet-facing connectivity. A common mistake is launching an ECS instance successfully but forgetting that the speed test tool or remote endpoint requires a public IP path and open ports.
Alibaba Cloud postpaid billing account Where to click and what to configure:
- Open the ECS instance details page
- Check whether a public IPv4 address has been assigned
- Open the security group settings attached to the instance
- Alibaba Cloud postpaid billing account Add inbound rules for SSH on port 22 for Linux or RDP on port 3389 for Windows
- If using
iperf3in server mode, allow the chosen test port such as 5201/TCP and optionally 5201/UDP - For HTTP/HTTPS file transfer testing, allow 80 and 443 if needed
Linux firewall commands after login:
sudo ufw allow 22/tcp
sudo ufw allow 5201/tcp
sudo ufw allow 5201/udp
sudo ufw enable
For firewalld-based systems:
sudo firewall-cmd --permanent --add-port=22/tcp
sudo firewall-cmd --permanent --add-port=5201/tcp
sudo firewall-cmd --permanent --add-port=5201/udp
sudo firewall-cmd --reload
Important considerations:
- Security group and OS firewall must both allow the traffic
- If you only open the Alibaba Cloud security group but block the port in the OS firewall, tests will still fail
- For safer testing, restrict source IPs to your office or home public IP instead of opening all addresses
Step 3: Connect to the server
For Linux, connect with SSH:
ssh root@your_public_ip
For Windows, use Remote Desktop Connection and enter the public IP plus the administrator username and password.
If you cannot connect:
- Check that the public IP is present
- Verify the security group rule direction and port number
- Confirm the instance status is Running
- Check whether the OS firewall is active and blocking the port
- For Windows, wait a few minutes after boot because RDP may not be available immediately
Step 4: Install speed testing tools
For Linux, two practical methods are recommended:
iperf3for controlled throughput tests between servers- Alibaba Cloud postpaid billing account
speedtest-clior the official Ookla CLI for internet speed checks to public test nodes
Ubuntu/Debian installation:
sudo apt update
sudo apt install -y iperf3 curl
If using Ookla speedtest CLI, follow the package instructions for your distribution. For a quick public test, some users install a CLI package from the vendor repository. For controlled comparisons, iperf3 is usually better.
Step 5: Measure download and upload speed correctly
There are three reliable testing patterns, and they answer different questions.
Method A: Test ECS internet speed to a public speed node
Alibaba Cloud postpaid billing account This is the simplest method when you want a quick view of internet download/upload throughput from the ECS instance.
Run the speed test tool from inside the ECS instance. The result reflects the path between Alibaba Cloud and the selected public test node, not the path to your home computer or your users in every region.
Use this method when:
- You want a fast initial check after purchase
- You need a rough comparison between two ECS regions
- You are validating whether public bandwidth is broadly functional
Be careful interpreting the result because public speed nodes have their own load and route limitations.
Method B: Test between two cloud servers with iperf3
Alibaba Cloud postpaid billing account This is the most useful method for migration planning and cross-region comparisons.
Alibaba Cloud postpaid billing account On the destination server, start the iperf3 server:
iperf3 -s
On the Alibaba Cloud ECS instance, run the client for a download-style test from the ECS perspective:
iperf3 -c remote_server_ip -P 4 -t 30
For reverse direction, which is often used to estimate upload-style behavior from ECS:
iperf3 -c remote_server_ip -P 4 -t 30 -R
Parameter notes:
| Command option | Meaning | Recommendation |
| -P 4 | 4 parallel streams | Use 4 to 8 streams to avoid underutilization on high-latency paths |
| -t 30 | 30-second test duration | Short tests can be noisy; 30 to 60 seconds is more stable |
| -R | Reverse direction | Useful to compare inbound versus outbound path behavior |
Important considerations:
- If you test over the internet, both sides must allow the
iperf3port - Single-stream tests may show lower throughput than the purchased bandwidth on long-distance routes
- CPU limits on small instances can reduce encrypted transfer rates if you test through application-layer tools
Method C: Real file transfer test
If your actual workload is website file delivery, backups, or media transfer, test with a real file. For example, upload a 1 GB test file via SCP, SFTP, or HTTPS and measure sustained speed. This reveals practical throughput including protocol overhead.
Example SCP upload from your local machine to ECS:
scp largefile.bin root@your_public_ip:/root/
Use this method when you care about real transfer behavior more than theoretical maximum throughput.
How to interpret the results without making the wrong purchasing decision
A low speed result does not automatically mean Alibaba Cloud ECS is the wrong choice. Check these factors first:
| Observation | Likely cause | What to verify |
| Result close to purchased bandwidth cap | Normal behavior | Bandwidth plan may be the limiting factor |
| Result much lower only from your home PC | Local ISP or Wi-Fi bottleneck | Test from another network or a second cloud server |
| Good speed within same country, poor cross-border speed | International routing or peering limitation | Try a different region closer to users or compare at another time |
| Very low speed on tiny instance | CPU bottleneck or insufficient parallelism | Increase vCPU or use multiple streams |
| Intermittent failure | Security group, firewall, or test node instability | Review logs and test against another endpoint |
If the result influences a purchasing decision, always test at least three times, at different times of day, and from more than one endpoint.
Alibaba Cloud postpaid billing account Configuration recommendations by scenario
For individual developers
If you only need to verify whether Alibaba Cloud is suitable for a personal site, VPN endpoint, or development host, choose pay-as-you-go, a modest 2 vCPU instance, Linux, and a region close to your primary audience. Do not subscribe first unless your tests are already complete. Keep the public bandwidth explicit and documented. Use SSH and iperf3 for testing, then release the instance if it is only for evaluation.
For enterprise users
Use a verified business account, assign RAM roles instead of using the root account for operations, and maintain a record of selected region, VPC, security group, and bandwidth values for auditability. For throughput validation, test from branch offices, other cloud environments, and your on-premises data center. Budget for snapshots if the server will be reused during repeated network comparison cycles.
Alibaba Cloud postpaid billing account For cross-border businesses
Create test ECS instances in at least two candidate regions and compare not just raw throughput but also latency and stability over a 24-hour window. A region with slightly lower peak speed but better consistency may be the better production choice. Expect additional compliance checks or payment review if the account origin, login IP, and payment source do not align cleanly.
For website applications
If the server will host a website, test both origin bandwidth and page delivery conditions. Open ports 80 and 443, configure DNS to a temporary hostname if needed, and compare direct origin access versus CDN acceleration if your audience is geographically distributed. A simple CLI speed test is not enough for web workloads; test actual file downloads and HTTPS responses.
For database workloads
Database traffic usually depends more on private network design, latency, and connection stability than public upload/download figures. If your application servers and databases are inside Alibaba Cloud, prioritize same-region VPC testing, internal connectivity, and packet stability over public internet speed. For migration windows, however, public or inter-region throughput tests using iperf3 remain important for estimating transfer duration.
Common troubleshooting during purchase and testing
Order cannot be completed
Check whether identity verification is unfinished, the payment method is unsupported, or the order has entered manual review. If you are using a new account, start with a small pay-as-you-go order rather than a large multi-instance purchase.
No public IP after instance creation
Recheck whether internet bandwidth or public IP assignment was enabled during purchase. Some users create a perfectly valid ECS instance inside a VPC but without public exposure, making internet speed tests impossible until a public internet configuration is attached.
SSH or RDP connection fails
Confirm the instance is running, the correct public IP is used, the security group allows 22 or 3389, and the OS firewall permits the same port. For Linux, use the console if SSH is blocked and adjust the firewall from there.
Speed is lower than expected
Test with multiple streams, a longer duration, another endpoint, and another time window. Also compare against a second cloud server instead of relying only on public speed nodes. If your purchased bandwidth is small, the result may already be normal.
Renewal risk for ongoing tests
If you plan recurring benchmark checks, set renewal reminders or auto-renew where appropriate. Losing the instance due to expiration can invalidate historical comparisons, especially if the public IP changes after recreation.
Optimization recommendations
Security optimization
- Limit SSH and RDP source IP ranges in security groups
- Close test ports like 5201 after benchmarking
- Use SSH keys instead of passwords for Linux where possible
- Keep snapshots before major reconfiguration if the instance will remain in service
Performance optimization
- Use
iperf3with parallel streams for long-distance testing - Choose a region close to your users instead of chasing a cheaper but distant region
- Avoid very small instance types when testing encrypted file transfers
- Run tests at multiple times to identify route congestion patterns
Alibaba Cloud postpaid billing account Cost optimization
- Use pay-as-you-go for short evaluations
- Do not overbuy CPU if your goal is only bandwidth validation
- Release temporary test instances after measurement
- Track public bandwidth charges carefully if running extended tests or repeated transfers
Stability optimization
- Record every parameter used in the test: region, instance type, bandwidth, endpoint, protocol, time window
- Use at least two test methods before making a migration decision
- Retest after any OS reinstall, network rule change, or bandwidth modification
- If comparing vendors or regions, keep the same file size, stream count, and test duration for consistency
The most reliable decision workflow is simple: complete verification first, use a valid payment path, create a small but properly networked ECS instance, test with both public and controlled methods, and only then decide whether to renew, scale, or migrate production traffic.

