Speed claims are easy to make. I wanted to see what HostArmada’s infrastructure actually delivers under real conditions, so I ran two separate sets of tests:
- A GTmetrix performance test on a live WordPress site hosted on the WP Speed Reaper shared plan,
- And a full server benchmark suite on the Cloud SSD VPS
The two tests answer different questions. The WordPress hosting test tells you what a real visitor experiences when they land on your site. The VPS benchmarks tell you what the underlying server is capable of at the infrastructure level.
Together, they give you a complete picture of what HostArmada’s network can do across two of its most popular product types.
Here is everything I found.
How I Tested
I used two distinct testing approaches, one for each product.
- For shared WordPress hosting, I signed up for the WP Speed Reaper plan, which runs on 6 CPU cores, 6GB RAM, and 40GB NVMe storage. Before running any tests, I installed content, including images, navigation menus, and multiple pages. Testing an empty site would not give a useful reading of real-world performance. I ran the GTmetrix test twice from the Frankfurt, Germany server location, which matches HostArmada’s primary data center, and used the most consistent result.
- For the Cloud SSD VPS, I deployed a test server on the Web Raider plan, running Enterprise Linux 9 on an AMD EPYC 7413 processor with 8GB RAM and 160GB NVMe storage, also hosted in Frankfurt. I ran five categories of server benchmarks using sysbench and stress-ng, covering raw CPU processing power, memory throughput, disk I/O speed, network performance, and system stability under sustained load.
1. Shared WordPress Hosting Speed Test
Test Configuration
- Plan: WP Speed Reaper ($3.95/mo promotional, $19.75/mo regular)
- CPU: 6 Cores
- RAM: 6GB
- Storage: 40GB NVMe SSD
- Sites: Unlimited
- Backups: 21 daily backups
- Security: Imunify360 WAF, malware scan and removal
- Caching: Dynamic caching included
- Test Tool: GTmetrix
- Test Location: Frankfurt, Germany
- Site Content: Full demo install with images, navigation, and multiple pages
GTmetrix Results
| Metric | Result |
| GTmetrix Grade | A |
| Performance Score | 100% |
| Structure Score | 96% |
| Time to First Byte (TTFB) | 24ms |
| First Contentful Paint | 392ms |
| Largest Contentful Paint (LCP) | 392ms |
| Total Blocking Time (TBT) | 34ms |
| Cumulative Layout Shift (CLS) | 0.01 |
| Fully Loaded Time | 514ms |
What These Numbers Mean
A 100% performance score on shared hosting with a fully built site running was not what I expected going in.
That result tells me the server, caching layer, and infrastructure are working together at a level most shared hosts do not reach.
TTFB: 24ms
This is the standout metric. TTFB measures how long it takes the server to respond to the browser’s first request. At 24ms, HostArmada responded almost instantly. Google considers anything under 200ms excellent for shared hosting. This result came in at roughly one-tenth of that threshold.
LCP: 392ms
The Largest Contentful Paint measures when the main content appears on screen. At 392ms, the page’s primary content was visible in under half a second.
Google’s benchmarks consider anything under 800ms exceptional and under 2.5 seconds good. HostArmada cleared the exceptional threshold with significant room to spare.
TBT: 34ms
Total Blocking Time measures how long the page was unresponsive during loading. At 34ms, the page stayed almost entirely interactive from the moment it started rendering. Visitors could click, scroll, and interact almost immediately.
CLS: 0.01
Cumulative Layout Shift measures how much the page jumps around while loading. A score of 0.01 means almost nothing moved unexpectedly. Google’s good threshold is under 0.1.
A high CLS score is one of the most frustrating experiences for real visitors and is a ranking signal Google factors into search positions.
Fully Loaded Time: 514ms
The entire page finished loading in just over half a second. For a site with real demo content, images, and navigation, that is an exceptional result on a shared plan.
What This Means for SEO
All three Core Web Vitals, LCP, TBT, and CLS, came in well inside Google’s good range. Google uses these scores as direct ranking signals, and a site that clears all three thresholds from day one starts with a technical SEO foundation that many sites spend months optimizing toward.
For anyone building a site that depends on organic search traffic, that starting point matters before you write a single word of content.
2. Cloud SSD VPS Server Benchmarks
Test Configuration
- Plan: Web Raider
- CPU: AMD EPYC 7413 24-Core Processor
- RAM: 8GB
- Storage: 160GB NVMe SSD
- OS: Enterprise Linux 9
- Location: Frankfurt, Germany
Full Benchmark Summary
| Benchmark | Result |
| CPU Events per Second | 1,591.42 |
| CPU Average Latency | 0.63ms |
| Memory Transfer Speed | 6,264 MiB/sec |
| Memory Operations per Second | 6,414,691 |
| Disk Read Speed | 77.83 MiB/s |
| Disk Write Speed | 51.89 MiB/s |
| Disk Average Latency | 0.05ms |
| Network Download Speed | 657.94 Mbit/s |
| Network Upload Speed | 550.87 Mbit/s |
| Network Ping | 2.344ms |
| Stress Test Bogo ops/s | 6,836.69 |
CPU Performance
I ran the sysbench CPU benchmark, which calculates prime numbers up to 20,000 to measure raw processing power. This simulates real workloads like PHP execution, concurrent user requests, and background jobs.
The server hit 1,591 events per second with an average latency of 0.63ms throughout the entire test. The worst spike reached only 1.68ms, and the gap between minimum and maximum latency was tiny. That consistency tells me the CPU was working at a steady pace from start to finish rather than peaking and dropping.
For practical use, a CPU at this level keeps WordPress responsive even with heavy plugins running and handles multiple concurrent visitors without slowing down.
Memory Performance
The sysbench memory benchmark measures how quickly RAM can read and write data. I ran it across 10GB of data in 1MB blocks, simulating database operations and application processes passing large datasets through memory.
Memory throughput hit 6,264 MiB/sec, and average latency was reported as 0.00ms throughout, meaning every operation completed faster than sysbench could measure. The worst spike across the full run was 0.45ms.
On a virtualised server, hitting these speeds tells me the KVM hypervisor is introducing very little overhead between the guest OS and physical memory.
Workloads like Redis caching, busy MySQL databases, and multiple simultaneous application processes will not create a bottleneck here.
Disk I/O Performance
I ran a random read and write test, which simulates how a web server actually accesses data: jumping between different locations on disk unpredictably, exactly what happens when multiple visitors are hitting your site at the same time.
| Metric | Result |
| Read Operations per Second | 4,981.26 |
| Write Operations per Second | 3,320.84 |
| Fsync Operations per Second | 10,630.59 |
| Read Throughput | 77.83 MiB/s |
| Write Throughput | 51.89 MiB/s |
| Average Latency | 0.05ms |
| Maximum Latency | 1.66ms |
The read speed of 77.83 MiB/s and write speed of 51.89 MiB/s are strong for a random mixed workload. Traditional SSD hosting typically struggles under this kind of test. NVMe storage handles random operations significantly better, and these results reflect that.
The fsync result of 10,630 operations per second is worth specific attention. A result above 10,000 indicates storage tuned for both data safety and speed, which matters for database-heavy applications where data integrity is critical.
Network Performance
I tested network speed using speedtest-cli, which automatically selected the best available server based on ping.
The test ran against a Hivelocity node in Frankfurt via Leaseweb Germany.
| Metric | Result |
| Download Speed | 657.94 Mbit/s |
| Upload Speed | 550.87 Mbit/s |
| Ping | 2.344ms |
Most VPS providers at this tier cap network ports at 500 Mbit/s. HostArmada came in above that threshold on both download and upload, which suggests the network is not being heavily shared across tenants.
The ping of 2.344ms to a Frankfurt node is well under the 10ms threshold that is considered excellent for a VPS.
At less than a quarter of that benchmark, the server responds to incoming requests almost instantly. That matters for real-time applications, API endpoints, and any service where response time is predictable.
Stress Test
I ran stress-ng across all 4 CPU cores and 2 memory stressors simultaneously for 5 minutes, simulating what happens during a sustained traffic spike where the server is running at full capacity for an extended period.
| Metric | Result |
| Test Duration | 300 seconds |
| CPU Bogo ops/s | 4,586.21 |
| VM Bogo ops/s | 9,768.60 |
| Total CPU Operations | 1,375,870 |
| Total VM Operations | 2,930,589 |
| Stressors Passed | 6 out of 6 |
| Failed Tests | 0 |
| Errors | 0 |
Every stressor passed. No failures, no crashes, no instability across the full five minutes.
The CPU completed 1,375,870 operations while the memory stressor ran alongside it at 9,768 bogo ops per second, and neither degraded during the run.
Budget providers often show strong results in short bursts by relying on CPU credits that deplete under sustained load. The consistency between the real-time and usr+sys scores here confirms the server was genuinely working at full capacity throughout, not coasting on burst performance.
Overall Speed Verdict
Across both products, HostArmada’s infrastructure held up well under real test conditions.
WordPress Hosting standout numbers:
- 24ms TTFB, roughly one-tenth of Google’s excellent threshold for WordPress hosting
- 100% GTmetrix performance score on a fully built site with real content
- 514ms fully loaded time with all three Core Web Vitals inside Google’s good range
The WP Speed Reaper plan sits well ahead of where most WordPress hosts land on the same tests. Any site hosted here starts with a strong technical SEO baseline before you touch a single plugin or optimization setting.
VPS standout numbers:
- 1,591 CPU events per second with consistent 0.63ms average latency
- 6,264 MiB/sec memory throughput with sub-microsecond average latency
- 657 Mbit/s download and 2.3ms ping, above the typical 500 Mbit/s VPS cap
- Stress test passed all six stressors across five full minutes with zero failures
For developers and agencies running production workloads, these numbers show a server that handles sustained pressure without degrading, not one that performs well in short bursts and drops off under load.
What the combined results tell you about HostArmada as a provider is that the speed claims hold up when tested. The consistency across two completely different product types, a WordPress plan and a self-managed VPS, points to infrastructure that performs reliably rather than selectively.
If speed is a priority and you want to test the infrastructure yourself, HostArmada backs WordPress hosting with a 45-day money-back guarantee and VPS plans with a 7-day guarantee.
