AI-assisted review
Instance benchmark

DigitalOcean s-4vcpu-8gb

The complete benchmark report. Every workload, in one place.

4 vCPU7.8 GiB RAMx86_64San Francisco, US$48.00 / month
Published report · 17 Oct 2026 · One deployment · 23 min test windowUbuntu 24.04.5 · Region sfo2 · Method pilot-1

All values below are medians of passed runs unless stated otherwise.

01 /

CPU performance

sysbench 1.0.20 prime workload · 1 thread and all 4 vCPUs.

342.5events/s
Single thread
1,389events/s
All 4 vCPUs
Events / sec
  • 1 thread
    342.5
  • 4 threads
    1,389

Median of 3 passed runs of about 30 seconds per thread count.

gzip fixed corpus · compression input throughput 160.8 MiB/s · single run.

02 /

Memory bandwidth

sysbench 1.0.20 memory · sequential read and write operations.

Read5,599 MiB/s
Write5,812 MiB/s
CopyNot measured
Throughput (MiB/s)
  • Read
    5,599
  • Write
    5,812
  • Copy
    Not measured

Median of 3 passed runs of about 30 seconds per operation. The copy workload is deferred in this methodology.

03 /

Disk performance

fio 3.36 · direct I/O · sequential, random and mixed jobs measured separately.

WorkloadThroughputp95 latency
Sequential read3,430 MB/s13,042 µs
Sequential write638.8 MB/s79,167 µs
Random read61,074 IOPS798.7 µs
Random write38,015 IOPS1,368 µs
Random mix · read31,271 IOPS1,286 µs
Random mix · write13,412 IOPS1,909 µs

Median of 3 passed runs of about 30 seconds per job. The mixed job's reads and writes are listed separately.

04 /

Network performance

Source: San Francisco, US · LibreSpeed CLI 1.0.14 · public HTTPS destinations.

DestinationDownloadUploadHTTP latencyHTTP jitterLoss
Frankfurt, Germany (Clouvider)
17/10/2026, 17:29:13 UTC · Separate VM network-only run · View reviewed run
479.9 Mbit/s145.6 Mbit/s145.2 ms0.09 msNot measured
New York, United States (2) (Clouvider)
17/10/2026, 17:29:13 UTC · Separate VM network-only run · View reviewed run
1,184 Mbit/s296.7 Mbit/s64.8 ms0.09 msNot measured

One 10-second download and upload per destination, three concurrent streams, IPv4 over HTTPS, with no configured rate limit. HTTP latency and jitter are application-level measurements; packet loss was not measured. These paths do not establish maximum link capacity.

Latest matching published internet speed test for this provider, plan, recorded region and resource class, taken on a separate VM. These values are not measurements from this full benchmark deployment. The original capped iperf path checks remain under Evidence; their values and test window are unchanged.

Browse all network speed tests.

12 GiB outbound-interface abort threshold is sampled; kernel buffering and sampling may overshoot. Retained counters disclose actual observed bytes.

05 /

PostgreSQL performance

PostgreSQL 16.15 · pgbench · in-memory and storage-bound datasets.

Dataset profileTransactions / secAverage latencyPassed runs
In memory1,7242.32 ms3 of 3 passed
Storage-bound1,5122.64 ms3 of 3 passed

Median of 3 passed runs of about 30 seconds per dataset. Latency percentiles were not measured.

Running PostgreSQL yourself? Read our PostgreSQL tutorials.

06 /

Consistency across runs

One deployment · repeated runs inside the recorded test window.

24-hour consistencyNot measured

Workload skipped · No measured score. This report covers a 23 min test window, so no time series exists and none is drawn.

RepetitionCPU 1 threadMemory readRandom readPostgreSQL
1359.4 events/s5,466 MiB/s61,074 IOPS1,654 tx/s
2342.1 events/s5,623 MiB/s65,546 IOPS1,724 tx/s
3342.5 events/s5,599 MiB/s56,888 IOPS1,914 tx/s

Individual runs, not medians. PostgreSQL uses the in-memory dataset. Spread within one test window does not establish long-term consistency.

07 /

Inspect the evidence.

Recorded environment, reviewed measurements and the published snapshot.

OSUbuntu 24.04.5 LTS
Architecturex86_64
Methodpilot-1
Deployments1
Window23 min
report.json

Reviewed, sanitized snapshot: every run, the environment record and all limitations. Raw intake logs remain private.

Read the full methodology

One deployment describes this report, not every server a provider operates.

A 23 min test window does not establish 24-hour consistency or long-term reliability.

AI-assisted review · Codex AI-assisted evidence review (user-delegated) · Reviewed 17 Oct 2026 · Published 17 Oct 2026 · Revision 1

Review and measurement limits
  • Single deployment; not an independent three-instance cohort.
  • No hardware attestation.
  • Pilot memory implements read/write; copy workload deferred.
  • Network endpoints must be supplied by the operator; no public third-party servers are used.
  • Database latency distribution is not available from this pilot parser; average latency only.
  • This profile is not a full implementation of all proposed methodology tests.
  • Resource-constrained environment: CPU workers and RAM reflect affinity/cgroup limits, not host capacity. Private resource-limits evidence retains both; do not merge container/VM or architecture cohorts.
  • Short profile: does not establish 24-hour consistency.
  • This report received AI-assisted evidence review under user-delegated authorization. This is not independent human verification or hardware attestation.
  • This pilot measures its recorded test window. It does not establish 24-hour consistency, long-term reliability or provider-wide performance.
  • Network tests use two controlled London endpoints, one stream, a 100 Mbit/s rate setting and a requested 64 MiB byte target per direction. Reverse-mode polling can exceed that byte target; actual sent and received bytes govern traffic accounting. These results do not measure maximum link capacity.
  • SKU and region were matched to retained deployment records. A missing detected SKU uses the operator-supplied plan and remains labelled supplied. Control-plane records and checksums do not attest the physical host or establish physical-host independence.
  • Package versions and available guard and journal evidence were reviewed. These records do not establish continuous absence of package activity throughout the test window.
  • The pilot includes memory read/write and average PostgreSQL latency. Memory-copy tests and PostgreSQL latency percentiles are not implemented. Skipped consistency and missing measurements are not scores.
Environment and configuration
Recorded fieldValue
ProviderDigitalOcean · detected
RegionSan Francisco, US (sfo2) · detected
Plans-4vcpu-8gb · supplied
List price$48.00 / month · $0.0714 / hour. List price on 17 Oct 2026, USD before tax. Source: DigitalOcean size catalog (API).
CPUDO-Regular
Architecturex86_64
Kernel6.8.0-142-generic
Virtualizationkvm
Agent0.2.0 (45388107250ca6e26b216f15a9c17b8d405ec1b7+source.ff98e866b26b3ac76b512554)
Dependenciessysbench: sysbench 1.0.20 · fio: fio-3.36 · iperf3: iperf 3.16 (cJSON 1.7.15) · postgresql: postgres (PostgreSQL) 16.15 (Ubuntu 16.15-0ubuntu0.24.04.1) · pgbench: pgbench (PostgreSQL) 16.15 (Ubuntu 16.15-0ubuntu0.24.04.1) · gzip: gzip 1.12 · apt:sysbench: 1.0.20+ds-6build2 · apt:fio: 3.36-1ubuntu0.1 · apt:iperf3: 3.16-1build2 · apt:postgresql-16: 16.15-0ubuntu0.24.04.1 · apt:postgresql-client-16: 16.15-0ubuntu0.24.04.1 · apt:gzip: 1.12-1ubuntu3.2
Configuration hash0284a3ee81606f2645108bbbc2006365e41a08e2b26be310958420bf7ef221f7
Started2026-10-17T12:05:48.224Z
Finished2026-10-17T12:28:37.803Z
Execution statuscompleted
Historical capped network path checks

Original full benchmark deployment · 17 Oct 2026. These controlled iperf path checks used a 100 Mbit/s rate setting and do not measure internet speed or maximum link capacity. RTT is the probe mean; jitter and packet loss were not measured. The original report JSON retains these measurements.

Controlled destinationDownloadUploadRTTJitterLoss
London endpoint 199.9 Mbit/s96.8 Mbit/s147.2 ms——
London endpoint 299.9 Mbit/s97.4 Mbit/s137.3 ms——
Every measured repetition and outcome

CPU performance

Workload / repetitionOutcomeMetricValue
CPU 1 thread(s) · repetition 1passedEvents per second359.4 events/s
CPU 4 thread(s) · repetition 1passedEvents per second1,430 events/s
CPU 1 thread(s) · repetition 2passedEvents per second342.1 events/s
CPU 4 thread(s) · repetition 2passedEvents per second1,280 events/s
CPU 1 thread(s) · repetition 3passedEvents per second342.5 events/s
CPU 4 thread(s) · repetition 3passedEvents per second1,389 events/s
gzip fixed corpus · repetition 1passedCompression input throughput160.8 MiB/s

Memory bandwidth

Workload / repetitionOutcomeMetricValue
Memory read · repetition 1passedBandwidth5,466 MiB/s
Memory write · repetition 1passedBandwidth5,911 MiB/s
Memory read · repetition 2passedBandwidth5,623 MiB/s
Memory write · repetition 2passedBandwidth5,812 MiB/s
Memory read · repetition 3passedBandwidth5,599 MiB/s
Memory write · repetition 3passedBandwidth5,730 MiB/s

Disk performance

Workload / repetitionOutcomeMetricValue
Disk write · repetition 1passedwrite bandwidth645.4 MB/s
Disk write · repetition 1passedwrite operations615.5 IOPS
Disk write · repetition 1passedwrite p50 latency45,875 µs
Disk write · repetition 1passedwrite p95 latency79,167 µs
Disk write · repetition 1passedwrite p99 latency221,250 µs
Disk read · repetition 1passedread bandwidth2,313 MB/s
Disk read · repetition 1passedread operations2,206 IOPS
Disk read · repetition 1passedread p50 latency10,289 µs
Disk read · repetition 1passedread p95 latency44,827 µs
Disk read · repetition 1passedread p99 latency67,633 µs
Disk randwrite · repetition 1passedwrite bandwidth146.1 MB/s
Disk randwrite · repetition 1passedwrite operations35,669 IOPS
Disk randwrite · repetition 1passedwrite p50 latency659.5 µs
Disk randwrite · repetition 1passedwrite p95 latency1,368 µs
Disk randwrite · repetition 1passedwrite p99 latency4,227 µs
Disk randread · repetition 1passedread bandwidth250.2 MB/s
Disk randread · repetition 1passedread operations61,074 IOPS
Disk randread · repetition 1passedread p50 latency469.0 µs
Disk randread · repetition 1passedread p95 latency798.7 µs
Disk randread · repetition 1passedread p99 latency1,581 µs
Disk randrw · repetition 1passedread bandwidth127.4 MB/s
Disk randrw · repetition 1passedread operations31,107 IOPS
Disk randrw · repetition 1passedread p50 latency561.2 µs
Disk randrw · repetition 1passedread p95 latency1,401 µs
Disk randrw · repetition 1passedread p99 latency3,686 µs
Disk randrw · repetition 1passedwrite bandwidth54.6 MB/s
Disk randrw · repetition 1passedwrite operations13,341 IOPS
Disk randrw · repetition 1passedwrite p50 latency518.1 µs
Disk randrw · repetition 1passedwrite p95 latency1,909 µs
Disk randrw · repetition 1passedwrite p99 latency3,621 µs
Disk write · repetition 2passedwrite bandwidth638.8 MB/s
Disk write · repetition 2passedwrite operations609.2 IOPS
Disk write · repetition 2passedwrite p50 latency44,827 µs
Disk write · repetition 2passedwrite p95 latency73,925 µs
Disk write · repetition 2passedwrite p99 latency325,059 µs
Disk read · repetition 2passedread bandwidth3,477 MB/s
Disk read · repetition 2passedread operations3,316 IOPS
Disk read · repetition 2passedread p50 latency9,241 µs
Disk read · repetition 2passedread p95 latency12,648 µs
Disk read · repetition 2passedread p99 latency16,908 µs
Disk randwrite · repetition 2passedwrite bandwidth200.8 MB/s
Disk randwrite · repetition 2passedwrite operations49,017 IOPS
Disk randwrite · repetition 2passedwrite p50 latency536.6 µs
Disk randwrite · repetition 2passedwrite p95 latency1,020 µs
Disk randwrite · repetition 2passedwrite p99 latency2,179 µs
Disk randread · repetition 2passedread bandwidth268.5 MB/s
Disk randread · repetition 2passedread operations65,546 IOPS
Disk randread · repetition 2passedread p50 latency428.0 µs
Disk randread · repetition 2passedread p95 latency733.2 µs
Disk randread · repetition 2passedread p99 latency1,679 µs
Disk randrw · repetition 2passedread bandwidth140.7 MB/s
Disk randrw · repetition 2passedread operations34,340 IOPS
Disk randrw · repetition 2passedread p50 latency485.4 µs
Disk randrw · repetition 2passedread p95 latency1,139 µs
Disk randrw · repetition 2passedread p99 latency3,523 µs
Disk randrw · repetition 2passedwrite bandwidth60.3 MB/s
Disk randrw · repetition 2passedwrite operations14,721 IOPS
Disk randrw · repetition 2passedwrite p50 latency452.6 µs
Disk randrw · repetition 2passedwrite p95 latency1,991 µs
Disk randrw · repetition 2passedwrite p99 latency3,457 µs
Disk write · repetition 3passedwrite bandwidth623.9 MB/s
Disk write · repetition 3passedwrite operations595.0 IOPS
Disk write · repetition 3passedwrite p50 latency46,399 µs
Disk write · repetition 3passedwrite p95 latency80,216 µs
Disk write · repetition 3passedwrite p99 latency308,281 µs
Disk read · repetition 3passedread bandwidth3,430 MB/s
Disk read · repetition 3passedread operations3,271 IOPS
Disk read · repetition 3passedread p50 latency9,241 µs
Disk read · repetition 3passedread p95 latency13,042 µs
Disk read · repetition 3passedread p99 latency18,743 µs
Disk randwrite · repetition 3passedwrite bandwidth155.7 MB/s
Disk randwrite · repetition 3passedwrite operations38,015 IOPS
Disk randwrite · repetition 3passedwrite p50 latency667.6 µs
Disk randwrite · repetition 3passedwrite p95 latency1,401 µs
Disk randwrite · repetition 3passedwrite p99 latency3,719 µs
Disk randread · repetition 3passedread bandwidth233.0 MB/s
Disk randread · repetition 3passedread operations56,888 IOPS
Disk randread · repetition 3passedread p50 latency485.4 µs
Disk randread · repetition 3passedread p95 latency880.6 µs
Disk randread · repetition 3passedread p99 latency2,408 µs
Disk randrw · repetition 3passedread bandwidth128.1 MB/s
Disk randrw · repetition 3passedread operations31,271 IOPS
Disk randrw · repetition 3passedread p50 latency561.2 µs
Disk randrw · repetition 3passedread p95 latency1,286 µs
Disk randrw · repetition 3passedread p99 latency3,588 µs
Disk randrw · repetition 3passedwrite bandwidth54.9 MB/s
Disk randrw · repetition 3passedwrite operations13,412 IOPS
Disk randrw · repetition 3passedwrite p50 latency510.0 µs
Disk randrw · repetition 3passedwrite p95 latency1,794 µs
Disk randrw · repetition 3passedwrite p99 latency3,457 µs

Network performance

Workload / repetitionOutcomeMetricValue
Network upload endpoint 1 · repetition 1passedSender throughput100.1 Mbit/s
Network upload endpoint 1 · repetition 1passedReceiver throughput96.8 Mbit/s
Network download endpoint 1 · repetition 1passedSender throughput100.0 Mbit/s
Network download endpoint 1 · repetition 1passedReceiver throughput99.9 Mbit/s
Latency endpoint 1 · repetition 1passedRound-trip mean147.2 ms
Network upload endpoint 2 · repetition 1passedSender throughput100.2 Mbit/s
Network upload endpoint 2 · repetition 1passedReceiver throughput97.4 Mbit/s
Network download endpoint 2 · repetition 1passedSender throughput100.1 Mbit/s
Network download endpoint 2 · repetition 1passedReceiver throughput99.9 Mbit/s
Latency endpoint 2 · repetition 1passedRound-trip mean137.3 ms

PostgreSQL performance

Workload / repetitionOutcomeMetricValue
PostgreSQL memory dataset · repetition 1passedTransactions1,654 tx/s
PostgreSQL memory dataset · repetition 1passedAverage latency2.41 ms
PostgreSQL memory dataset · repetition 2passedTransactions1,724 tx/s
PostgreSQL memory dataset · repetition 2passedAverage latency2.32 ms
PostgreSQL memory dataset · repetition 3passedTransactions1,914 tx/s
PostgreSQL memory dataset · repetition 3passedAverage latency2.09 ms
PostgreSQL storage dataset · repetition 1passedTransactions1,512 tx/s
PostgreSQL storage dataset · repetition 1passedAverage latency2.64 ms
PostgreSQL storage dataset · repetition 2passedTransactions1,444 tx/s
PostgreSQL storage dataset · repetition 2passedAverage latency2.77 ms
PostgreSQL storage dataset · repetition 3passedTransactions1,557 tx/s
PostgreSQL storage dataset · repetition 3passedAverage latency2.57 ms

Consistency over time

Workload / repetitionOutcomeMetricValue
24-hour consistency · repetition 1skippedNo measured scoreNot measured