Best Value Cloud Storage: Performance & Cost Comparison of Top Providers

Overview

Cloud storage has become one of the hottest products in the cloud computing field. It offers on‑demand usage, pay‑as‑you‑go pricing, no extra hardware, and no dedicated maintenance staff. Because of these advantages, many cloud storage providers have emerged, making selection difficult. This article provides an objective performance evaluation to help choose the right provider.

Cloud storage types include object storage (cold and hot), block storage, and table storage. This test focuses on hot object storage, comparing the cost‑performance of five vendors: AWS, Azure, Qiniu, Huawei Cloud, and Alibaba Cloud.

Test Tools and Methods

The performance test uses the open‑source tool Apache JMeter 2.13 . The method calls each vendor’s latest Java SDK to perform upload (PUT) and download (GET) operations on objects of sizes 1 KB, 10 KB, 100 KB, 1 MB, 10 MB, and 1 GB. The test environment uses a VM located in the same region as the storage service, with 1000 Mbps bandwidth, CPU utilization < 50 %, bandwidth usage < 70 %, and disk I/O < 80 MB/s.

Performance metrics:

Response Time (RT): 50th percentile response time. Transactions Per Second (TPS): average number of transactions per second.

Requests: GET (download) and PUT (upload).

Pricing data: Official prices as of November 2015.

Performance Test Results

Approximately 30 million objects of six different sizes were pre‑loaded into each provider’s storage. Random object keys were used for downloads to avoid cache effects. Three threads (one for 1 GB files) were used, with thinking time to keep the test machine from becoming a bottleneck.

Download (GET) results

Small files (1 KB, 10 KB) performed better on AWS and Azure, while large files (1 MB, 10 MB) showed superior performance on Qiniu and Alibaba Cloud.

Upload (PUT) results

Minimize data transfer hops.

When writing to multiple nodes, send data from the write node to an intermediate node before distributing further to avoid bandwidth limits.

For small files (1 KB, 10 KB), Alibaba Cloud and Qiniu had the best response times. For larger files (1 MB, 10 MB, 1 GB), AWS performed well, while Azure ranked lowest and timed out on 1 GB uploads, requiring multipart upload.

Cost Simulation (Based on User Scenario)

A typical mid‑size customer scenario: 10 TB storage, 100 GB daily download traffic, 500 k GET/PUT requests per day. Prices (Nov 2015) were collected from each vendor. Azure was the most expensive, Qiniu the cheapest (about 2.5× cheaper than Azure).

Cost‑Performance Analysis

Cost‑performance is calculated as performance relative value divided by price relative value. Performance relative value combines RT (weighted 60 %) and TPS (weighted 40 %). Using 1 KB and 10 MB files as benchmarks and Alibaba Cloud as the baseline (value = 1), the ranking is:

Qiniu

Alibaba Cloud

AWS

Huawei Cloud

Azure (lowest)

<code>CostPerformance = PerformanceRelativeValue / PriceRelativeValue

PerformanceRelativeValue = (RTRelativeValue / 4 * 60%) + (TPSRelativeValue / 4 * 40%)

RTRelativeValue = Alibaba/AWS (GET 1K RT) + Alibaba/AWS (PUT 1K RT) + Alibaba/AWS (GET 1M RT) + Alibaba/AWS (PUT 1M RT)

TPSRelativeValue = AWS/Alibaba (GET 1K TPS) + AWS/Alibaba (PUT 1K TPS) + AWS/Alibaba (GET 1M TPS) + AWS/Alibaba (PUT 1M TPS)

PriceRelativeValue = AWS price / Alibaba price</code>

Note: Smaller RT is better (use Alibaba/AWS ratio), larger TPS is better (use AWS/Alibaba ratio). Values are based on the 50th percentile.

Conclusion

For startups and enterprises transitioning to cloud services, selecting the right cloud storage provider is crucial. Performance‑to‑price ratio provides a solid basis for comparison, complemented by considerations of usability, service quality, security, and regional coverage.