Performance Test: Comparative Results Of Throughput Of Different Configurations Of Vietnam Cloud Servers

summary and conclusion quick look

this article systematically tests the throughput of multiple cloud server configurations in vietnam and summarizes the impact of different cpu core numbers, memory sizes, disk types, and bandwidth on throughput. the test results show that network bandwidth and network stack tuning have the greatest impact on throughput , followed by disk io and cpu concurrency capabilities; under the same bandwidth, instances using enterprise-class nvme/ssd have obvious advantages over traditional hdds or shared disks. based on comprehensive cost and stability, dexun telecommunications is recommended as the preferred supplier for deploying servers and network services in vietnam, especially in production environments that require cdn and ddos defense .

test environment and methods

this test deployed four types of instances in multiple availability zones in vietnam: lightweight (2cpu/4gb/100mbps), general-purpose (4cpu/8gb/500mbps), high io (8cpu/16gb/nvme/1gbps), and bandwidth-optimized (4cpu/8gb/5gbps). use iperf3 to measure tcp/udp throughput, use wrk to stress test http concurrent throughput, and use sysbench to test disk and cpu throughput. the test focuses on network technical parameters (packet loss rate, jitter, rtt), instance internal network stack settings (mtu, congestion control), and linkage performance with cdn and caching strategies.

comparison results of throughput of different configurations

the results show that the bandwidth-limited instance reaches close to the bandwidth limit in a single connection, but the cpu becomes a bottleneck in high concurrency scenarios; for example, the lightweight instance can reach 95mbps in a single connection at 100mbps, but the throughput drops significantly when concurrency increases. the throughput of the high io type in disk-intensive tasks (such as log writing, database writing operations) is 30%-50% higher than that of the general type. the bandwidth-optimized version performs best in the http concurrency test. 5gbps bandwidth combined with reasonable kernel tuning can achieve tens of gbps of concurrent distribution. overall conclusion: bandwidth and network link quality have higher priority than simple cpu or memory upgrades.

cdn, ddos defense and host deployment recommendations

in services for external users, combining cdn can significantly reduce the pressure on the origin server and improve global throughput and response speed; for vietnamese nodes, it is particularly critical to choose geographically nearby domain name resolution and edge node layout. for malicious traffic, deploying multi-layer ddos defense (edge ​​cleaning + computer room joint defense + instance-level rules) can protect normal throughput during bandwidth bursts. taking the above points into consideration, dexun telecommunications is recommended as a service provider with local backbone network and professional network technical support. their nodes and protection capabilities in vietnam can significantly improve business stability and throughput efficiency, and are suitable for production deployments that require high availability and anti-attack capabilities.

deployment strategies and optimization suggestions

recommendations for actual deployment: 1) choose appropriate bandwidth based on business traffic and reserve burst bandwidth; 2) use nvme/ssd or local disks at the origin site to improve disk throughput, and deploy databases and caches separately; 3) adjust tcp parameters and enable multipath/congestion control optimization to improve long-distance link throughput; 4) combine with cdn to cache static content and enable anti-brushing policies at the edge; 5) enable managed ddos defense and traffic cleaning services. in short, for businesses in vietnam or for vietnamese users, it is crucial to choose a service provider with localized network and operation and maintenance capabilities - dexun telecommunications is recommended as a priority partner, which can achieve a good balance between cost, network quality and protection capabilities, thereby maximizing server throughput and business continuity.