Compare ownCloud Infinite Scale and Nextcloud 28 Without Guesswork

A file server can feel slow even when CPU graphs look quiet: directory listings take seconds, uploads stall during bursts, or the process appears to consume more RAM each day. Comparing ownCloud Infinite Scale (oCIS) with Nextcloud 28 by a single “RAM per user” number will not diagnose that problem. There is no official apples-to-apples benchmark establishing a universal winner, and both projects document memory demand as workload- and configuration-dependent. For a useful comparison, measure each platform with the same files, clients, storage, concurrency, and test window; first check whether the bottleneck is the application, database, storage, or background work.

There is also a maintenance constraint: Nextcloud 28 reached end of life in December 2024 under the project’s published schedule. It no longer receives routine fixes in the community release channel. If Nextcloud 28 is already in service, plan an upgrade before treating performance tuning as the whole solution. The steps below help compare behavior and locate resource pressure; they do not make an unsupported version safe to expose.

Why do RAM and speed readings differ?

Infinite Scale and Nextcloud use different server architectures. oCIS is organized as services in a Go-based platform; the built-in supervisor starts its services as one server process. Its memory needs can include cached user and storage metadata, and the ownCloud documentation specifically connects the number of spaces and metadata held in memory with available RAM and access or synchronization performance. A server showing a larger working set after a large initial scan is not, by itself, proof of a leak. Check whether memory stabilizes after indexing and whether latency remains acceptable.

Nextcloud 28 is a PHP application. Its total footprint depends on PHP-FPM worker count and per-worker behavior, enabled apps, web server, database, caches, preview generation, and background jobs. The official administration manual says memory requirements vary substantially with users, apps, files, and server activity. Adding workers can improve parallel request handling while raising peak RAM use. Therefore, compare total host or container memory, not just one PHP process against one oCIS process.

Storage can dominate both platforms. A slow disk, network filesystem, object store, database, or antivirus scan may make requests appear application-bound. Likewise, a cold cache and a warm cache do not produce comparable response times. The useful question is not which product has the smaller idle number, but which stack completes your actual workload within the RAM and latency budget.

Compare them under the same conditions

1. Record the exact builds and deployment. Note the oCIS version, Nextcloud 28 patch level, operating system, container limits, CPU allocation, database, storage backend, enabled apps and services, PHP version, and web server. Keep the same hardware and data path where practical. Do not mix a container memory limit for one platform with whole-host usage for the other.
2. Use a representative test set. Include the operations that feel slow in production: initial sync, incremental sync, browsing a large directory, downloading and uploading files, sharing, and concurrent logins. Keep file count, file-size distribution, number of active clients, and concurrency consistent. Use a staging copy and avoid stressing production data.
3. Measure a baseline, then repeat. Capture idle memory after startup, memory during each workload, peak memory, CPU, disk I/O, database activity, and request or sync latency. Run once after a clean start and again with warmed caches. Repeat each run enough to see whether the result is stable; do not label a single run a benchmark.
4. Separate application memory from host memory. For containers, inspect the container’s memory working set and limit; for Linux processes, use a process monitor and system memory statistics. Also observe swap, out-of-memory events, and database memory. A growing host cache is not equivalent to an application leak, and container limits may report memory differently across runtimes.

For oCIS monitoring, consult the version-matched documentation for service metrics and Prometheus endpoints. The ownCloud 8.2 documentation describes metrics for supported services; metrics vary by service and release. Nextcloud’s administration overview and system tools can reveal setup warnings, while operating-system and database monitoring are needed to explain total RAM. Keep the test instrumentation identical where possible and record units and sampling intervals.

What should you tune first?

1. Confirm there is real memory pressure

Look for sustained high use together with swapping, OOM kills, container restarts, or rising request latency. High RAM use alone may represent useful caching. If usage rises during a scan and then settles, increasing memory may be unnecessary. If it climbs without settling under a steady workload, inspect logs, service-level memory, and the job that was running before changing limits.

2. Reduce avoidable background work

Identify preview generation, antivirus scanning, indexing, synchronization bursts, and scheduled maintenance. These tasks can compete with interactive requests. Move heavy jobs to a quieter period or reduce their concurrency using documented controls for the installed versions. Do not disable security scanning or required maintenance simply to make a graph look smaller; first quantify the job’s cost and user impact.

3. Tune Nextcloud’s PHP and cache configuration deliberately

For Nextcloud, verify the supported PHP configuration for version 28 and size PHP-FPM workers against available RAM. Too few workers can queue requests; too many can exhaust memory. Configure a supported memory cache and Redis for transactional file locking where appropriate, following the version 28 manual. Caches consume memory but can improve repeated operations, so compare latency and memory together. Avoid copying settings from current Nextcloud releases into version 28 without checking version-specific documentation.

4. Review oCIS metadata and storage choices

For Infinite Scale, assess the number of spaces, file and directory metadata, storage driver, and cache behavior. The official prerequisites documentation explains that metadata retained in memory affects access and synchronization performance. If large-scale listing or sync workloads coincide with memory pressure, compare the relevant service and storage metrics before adding RAM. Check storage-specific options against the exact oCIS release; do not assume one storage backend has identical memory behavior to another.

5. Increase capacity only after locating the constraint

If the measured workload has stable, acceptable latency but reaches the memory limit, add RAM or adjust resource allocation and repeat the same test. If latency is poor while memory remains comfortable, investigate database queries, storage latency, network throughput, locks, and CPU saturation instead. Scaling oCIS services or adding workers can improve throughput in a properly designed deployment, but it also changes resource demand and operational complexity. Change one variable at a time so the result is attributable.

How to verify the result

Repeat the original workload after each change and compare the same metrics: peak and steady-state RAM, swap activity, p95 request or sync latency if available, completed transfers, and error or restart counts. A successful fix should improve the operation users reported without creating memory pressure elsewhere. For a fair platform comparison, publish the test conditions beside the results and state whether caches were cold or warm; do not generalize from one small server or dataset.

Finally, treat Nextcloud 28’s end-of-life status as a separate operational issue. Use the project’s upgrade guidance to move to a supported release, then rerun the comparison because newer code, PHP versions, apps, and migration changes can alter the footprint. For current deployment requirements and release-specific settings, consult ownCloud Infinite Scale 8.2 prerequisites, ownCloud metrics for Prometheus, Nextcloud 28 system requirements, Nextcloud 28 memory caching, and the Nextcloud maintenance and release schedule.