Quick Summary (TL;DR): Upgrade the Plex/Jellyfin server only when you have proved that video transcoding or application compute is the bottleneck. If playback is Direct Play and the server is largely idle, a faster CPU or NASAliExpress price will rarely cure buffering: investigate the client, Wi-Fi, 100Mb/s Ethernet port, network path and source bitrate first. Upgrade NASAliExpress price storage when you need capacity, have disk-health problems, or have demonstrated sustained storage latency—not because a 4K file sounds demanding. Move media-server metadata to an SSD if poster browsing and library scans are slow. A modern Intel Quick Sync mini PCAliExpress price plus an existing NASAliExpress price is often the least disruptive fix for genuine transcoding limits; a NASAliExpress price-only replacement makes sense when you also need new bays, software or storage features. The decision should follow a repeatable playback test, not a CPU benchmark or a claim about how many “4K streams” a box can run.
Plex Server vs NAS Storage: The Upgrade Decision in One Table
| What you actually observe | Most likely place to investigate | Free or low-cost check | When an upgrade is justified |
|---|---|---|---|
| One particular TV buffers; phone or PC plays the same file | Client codec support, TV network port, Wi-Fi | Try a capable wired client and compare Direct Play status | Client/connection before server |
| Remote stream forces a 4K-to-1080p conversion; CPU stays busy | Video transcoding hardware, software settings, upload limit | Inspect Plex Dashboard and hardware-transcode indicator | Compute only if conversion remains necessary |
| All clients pause while the NAS reports disk errors or very high latency | NAS drive health, array rebuild, I/O contention | Check storage alerts, SMART health, rebuild/scrub and workload | Repair or replace faulty/overloaded storage |
| Playback is fine but posters and searches are sluggish | Media-server database and thumbnail storage | Compare app-data location and disk activity | SSD for app data; not necessarily SSD for films |
| A 100Mb/s TV Ethernet port or congested Wi-Fi struggles with high-bitrate files | Client-side network | Test another path/client; check negotiated link rate | Better client, Wi-Fi or networking, not a new NAS |
| NAS is running out of usable capacity | Capacity and backup design | Check actual used space, free space, expansion rules | More/larger bays and drives, plus backup capacity |
| Many simultaneous streams and other jobs saturate a shared 1GbE link | Aggregate LAN throughput | Check concurrent traffic and actual link speed | 2.5GbE/10GbE only where both ends and switch support it |
The word server is overloaded. Plex Media Server may run on a NAS, a mini PC, a desktop or a virtualised host, while the video files live on a separate NAS. In that two-box arrangement the NAS reads the files, the media server decides whether to send or convert them, and the playback client must receive and decode the result. A problem at any of those points can look like “the NAS is too slow”.
First Identify Which Box Is Doing Which Job
Draw the actual path before spending anything. In a typical separated setup it is NAS disks → NAS Ethernet → switch/router → mini PC running Plex/Jellyfin → switch/Wi-Fi → TV or phone. With Plex installed on the NAS, the storage and compute stages share one enclosure, but the playback client and network remain independent bottlenecks. For remote viewing, add the home internet upload link and the remote download link to the chain.
Write down the media server host, the location of the film files, whether the client is wired or wireless, and whether playback is local or remote. If you do not know which host actually runs Plex, check its server settings or the container/host management page rather than assuming it is the NAS.
The 15-Minute Test That Should Come Before Any Purchase
- Choose one reproducible problem file. Record its video codec (H.264, HEVC, AV1, etc.), resolution, reported bitrate, audio track and subtitle type. Note whether it is a high-bitrate remux or a more compressed encode.
- Reproduce the problem on the original client. Note whether it is local or remote, the selected quality, and when buffering begins. A failure only during seeking may point to a different issue from steady-state playback.
- Inspect the media-server dashboard during playback. In Plex, expand the Now Playing details and note Direct Play, Direct Stream or Transcode for video and audio. If it is transcoding, check whether video shows
(hw). Jellyfin shows playback method and transcoding reasons in its dashboard/logs. - Observe the relevant host while the issue happens. Check CPU utilisation per core, GPU video engine where exposed, RAM pressure, disk latency/health and network throughput. A single saturated thread may matter even when total CPU looks modest.
- Change one variable. Try another client on wired Ethernet, disable a problematic subtitle track, or temporarily select a compatible audio track. For remote playback, repeat locally. Keep the file and server unchanged.
- Record what changed. If the same file Direct Plays on a different client, do not immediately replace the NAS. If a necessary video transcode cannot keep up despite correctly configured acceleration, then investigate compute.
Do not change five settings at once. The purpose is to isolate the stage that fails, not to make the problem disappear temporarily without understanding why.
Direct Play, Direct Stream and Transcode Are Different Workloads
| Playback path | What changes | Typical server workload | First upgrade to consider |
|---|---|---|---|
| Direct Play | Original container, video and audio are delivered without conversion | Very little media processing; mainly file I/O and networking | Usually none: investigate client/network first |
| Direct Stream / remux | Container is repackaged; video often remains untouched | Low processing compared with video re-encoding | Check client/container compatibility |
| Audio-only conversion | Video stays original; audio is re-encoded | Often modest CPU work; still inspect symptoms | Compatible client/audio track |
| Video transcode | Video is decoded and re-encoded, perhaps resized or tone-mapped | Potentially heavy CPU/GPU/video-engine workload | Hardware acceleration or compute if genuinely required |
| Subtitle burn-in | Subtitle pixels are composited into video | Can force full video transcoding and reduce throughput | Try compatible text subtitles or another client |
Plex documents the distinction between Direct Play and Direct Stream; Jellyfin distinguishes Direct Play, Remux, Direct Stream and video Transcode in its playback documentation. A device may support the video codec but not the audio format, subtitle type or file container. It is the combination that determines whether the server must convert anything.
A useful experiment is to play the same troublesome film with subtitles off, then with a plain SRT track, then with an image-based PGS track if available. If only the last case forces a video transcode, replacing the storage array is not the right response. Similarly, a browser client and a dedicated media player can have very different codec capabilities.
How to Tell Whether the Plex CPU or GPU Is the Problem
In Plex Web, open Activity → Dashboard, expand the Now Playing details and inspect the video conversion path. Plex documents (hw) next to video when hardware acceleration is actually in use. A Plex Pass is required for Plex hardware-accelerated streaming; owning an Intel iGPU does not automatically enable the feature. Jellyfin does not require a paid licence for hardware acceleration, but the GPU, drivers and server configuration still have to support the requested codec pipeline.
If a conversion is occurring, look for the reason: a remote bandwidth limit, unsupported codec, client quality setting, incompatible subtitle or audio track, or HDR-to-SDR requirement. Do not infer the reason merely from the resolution of the original file. A modern TV playing an HEVC file directly may use less server CPU than a low-resolution file that must be transcoded into a different codec.
Hardware acceleration: verify the actual pipeline
- Plex: verify Plex Pass, Settings → Server → Transcoder → Show Advanced → Use hardware acceleration when available, then the
(hw)playback indicator. Plex currently labels HEVC output encoding experimental; do not assume every client or device will use it. - Jellyfin: choose the appropriate hardware-acceleration backend, such as Intel QSV/VA-API or another documented supported method, and confirm the server can access the hardware. Review the playback method and FFmpeg log when conversion fails.
- Docker: confirm the container can access the host graphics device and permissions (commonly
/dev/drion Intel Linux systems). A Quick Sync-capable CPU is irrelevant to the container if the video device is not exposed. - Virtual machines: verify actual hardware passthrough and driver support rather than assuming that the host iGPU is automatically available to a guest.
- HDR and subtitles: test the precise HDR tone-mapping and subtitle burn-in path you use. “Can play 4K” is not a reproducible transcoding benchmark.
If you see sustained CPU saturation and the server cannot convert the file faster than real time, a compute upgrade may be appropriate. But first confirm that conversion is necessary and that a supported video engine is being used. Changing the client or playback settings can sometimes remove the transcoding requirement altogether.
When a Mini PC Upgrade Beats Replacing the NAS
A separate media-server mini PC is attractive when the NAS is healthy, has enough drive bays and serves files reliably, but its built-in processor is inadequate for the video conversions you genuinely need. You can keep the existing disks and NAS operating system, move Plex/Jellyfin compute to the mini PC, and mount the NAS share over a stable wired connection. This is an architectural change, not a guarantee of faster playback in every case.
Choose the compute host by verified codec encode/decode support, software support, RAM and storage layout, not simply CPU benchmark scores. For Intel-based systems, the exact Quick Sync generation matters. An N100/N150-class mini PC can be sensible for a modest home media workload, but do not promise a fixed number of concurrent 4K transcodes: source bitrate, codec, HDR processing, subtitles, application version and thermal limits change the result. A higher-end Core/Ryzen system may be justified for additional VMs or workloads; it is not automatically better for media if the relevant fixed-function video engine is absent or unsupported.
The trade-offs of a second box are extra idle electricity, another power supply, patching and backups for its application database, and a new dependency on NAS-share availability. Keep the Plex/Jellyfin database and thumbnail cache on the mini PC’s local SSD when practical; leave the large film files on NAS HDDs. Ensure the NAS share mounts reliably before the media server scans the library, and preserve stable paths and permissions.
When the NAS or Its Disks Really Need Upgrading
A NAS replacement or storage expansion is justified by capacity, reliability, required bays, storage features or measured I/O constraints. It is not automatically justified by an isolated playback buffer. Check the following before buying:
- Capacity: the pool is close to its platform-recommended free-space threshold and you need room for new media, snapshots or backups. Account for usable RAID capacity rather than raw disk labels.
- Health: SMART alerts, unreadable sectors, degraded arrays, failing disks, repeated disconnects or excessive rebuild risk. Replace failed components and validate backups before considering performance upgrades.
- Contention: playback interruptions correlate with backup jobs, scrubs, array rebuilds, large downloads or multiple simultaneous users. Scheduling and workload separation may be cheaper than new disks.
- Measured latency: a disk or pool shows sustained high latency and insufficient read throughput under the real workload, after network and compute are ruled out.
- Features: you require more bays, snapshots, encrypted shares, better backups, or supported applications unavailable on the existing NAS.
For a normal sequential film read, even a mechanical HDD generally has substantial throughput relative to one compressed video stream. RAID 5, RAID 6, SHR and other layouts change fault tolerance and capacity, not the basic need to diagnose where a delay occurs. RAID is not a backup. Budget for a second independent copy when expanding storage, and test a restore before retiring the old system.
SSD Cache, SSD Media Library or SSD Application Data?
| Where the SSD goes | Likely benefit | What it usually will not fix |
|---|---|---|
| Plex/Jellyfin app data (database, posters, thumbnails) | Faster random reads, library browsing and some scans; avoids competing with HDD media I/O | Insufficient video encode/decode performance |
| Transcode temporary directory | Can help if the existing temporary disk is measurably slow or full; requires space and endurance planning | A video engine that cannot transcode in real time |
| NAS SSD cache | May help repeated random-I/O workloads, depending on NAS implementation | A single sequential film stream that already reads comfortably from HDD |
| Entire film library on SSD | Low latency, silence and high aggregate I/O at greater cost per TB | Client codec incompatibility, slow Wi-Fi, remote upload limitations |
If the complaint is “posters take ages to load” but actual playback is smooth, an SSD for the media server’s application data is often the sensible first storage experiment. Back up the application database before moving it, follow the media-server vendor’s migration procedure, and avoid blindly copying an active database while it is being written.
Do the Bandwidth Maths Before Buying 2.5GbE or 10GbE
Video bitrates are commonly shown in megabits per second (Mb/s), while file transfers often show megabytes per second (MB/s). Divide Mb/s by eight to convert to MB/s. The link speeds below are theoretical line rates, not guaranteed application throughput:
| Reported video bitrate | Equivalent average data rate | Interpretation |
|---|---|---|
| 20 Mb/s | 2.5 MB/s | Comfortably below the nominal bandwidth of ordinary wired Ethernet |
| 50 Mb/s | 6.25 MB/s | Usually manageable on a healthy 100Mb/s link, but leave room for peaks |
| 80 Mb/s | 10 MB/s | Can approach the practical limits of a 100Mb/s client port when peaks and overhead are included |
| 120 Mb/s | 15 MB/s | Exceeds the 12.5MB/s theoretical ceiling of 100Mb/s Ethernet |
| 200 Mb/s | 25 MB/s | Well below 1GbE line rate, but demands a capable client/network path |
For reference, 100Mb/s = 12.5MB/s, 1GbE = 125MB/s, 2.5GbE = 312.5MB/s and 10GbE = 1,250MB/s before protocol overhead. These are arithmetic conversions, not benchmark measurements. Real throughput is lower; Wi-Fi performance is especially variable. A file’s average bitrate also hides peaks, so do not size a marginal connection to exactly its reported average.
A single 80Mb/s film is roughly 10MB/s on average. That is not a compelling reason to replace a functioning 1GbE NAS. It is a reason to investigate a TV with a 100Mb/s wired port, poor Wi-Fi or a busy home network. For several simultaneous high-bitrate clients plus backups, aggregate traffic can eventually justify faster NAS and server links. Every segment—NAS port, server NIC, switch, cable and relevant client path—must support the intended speed.
A quick, safe network isolation test
On a trusted local network, iperf3 can measure throughput between two compatible computers without reading the NAS disks. It is useful for separating a network limitation from storage. Install it from your operating system’s trusted package source, then run:
# On the receiving test computer:
iperf3 -s
# On the sending test computer (replace with the receiver LAN IP):
iperf3 -c 192.168.1.50 -t 20Run the test in the same direction as the actual media flow where possible. Do not expose the test server to the internet, and stop it after the test. A healthy iperf3 result does not prove that the NAS disks or the TV decoder are fast; it only narrows the problem. For storage, compare a large file copy from the same NAS share to a capable wired computer, noting caching, concurrent tasks and protocol overhead. Avoid destructive storage benchmarks on a live array.
Local Playback and Remote Playback Need Different Diagnoses
Local buffering: verify the client’s actual negotiated link speed, Wi-Fi signal/interference, codec compatibility, source-file peaks and playback method. The internet connection normally should not carry a direct LAN media stream. A TV with 100Mb/s Ethernet may perform worse with a high-bitrate remux than a modern streaming box connected over good Wi-Fi or Gigabit Ethernet.
Remote buffering: the home connection’s upload speed becomes part of the media path. Plex documents upload-speed and remote-quality controls, some of which require Plex Pass. If the upload link cannot sustain the source bitrate, the server may need to transcode to a lower bitrate. A faster NAS network port cannot increase the internet uplink. Check whether the stream is direct or indirect/relayed, confirm the remote client’s quality settings and test the actual upload connection at a representative time.
A common expensive mistake is replacing the NAS because remote playback is poor, only to discover that the internet upload is the bottleneck. Another is buying a faster CPU for a local playback issue caused by a television’s codec or subtitle limitations. Both are avoidable with the same controlled test.
Four Realistic Upgrade Scenarios
Scenario 1: One 4K remux buffers on the living-room TV
Plex reports Direct Play and the server CPU is mostly idle. The same file plays smoothly on a wired desktop. The TV uses a 100Mb/s port or unstable Wi-Fi. Do not replace the NAS or CPU first. Try a better playback client, a suitable network path, or a compatible lower-bitrate version of the file. Check peaks rather than relying on average bitrate alone.
Scenario 2: Remote viewers force video conversion
A 4K HEVC/HDR source is converted to 1080p H.264 because of remote bandwidth or client support. The server cannot keep up, and the Plex dashboard shows no usable hardware acceleration. First verify the Plex Pass requirement, transcoder settings, hardware capability and container GPU access. If the platform cannot accelerate the necessary pipeline and conversion cannot be avoided, upgrade compute; the existing NAS disks may be entirely adequate.
Scenario 3: Films play fine, but the interface is painfully slow
Movie playback is stable, but posters, search and library navigation are sluggish while a large HDD pool handles thumbnail/database activity. Back up the application state, then evaluate moving Plex/Jellyfin app data to SSD. Do not assume that a new NAS CPU or all-SSD media library is required. Check RAM pressure and application database health too.
Scenario 4: Playback pauses during backups and array maintenance
The NAS reports heavy disk activity during scheduled backup jobs or an array rebuild. First move background work outside viewing hours, check drive health, and test whether playback recovers. If I/O pressure is a persistent requirement from many simultaneous users, then faster disks, a different pool layout or a separate storage workload may be justified. Capacity and resilience still need independent backup planning.
One NAS or Separate NAS and Media Server?
| Design | Strongest reason to choose it | What you take on | Best fit |
|---|---|---|---|
| Plex/Jellyfin on NAS | One enclosure and simpler management; local access to files | Storage and app workloads share CPU/RAM and maintenance windows | NAS already has compatible media engine and enough bays |
| Mini PC running Plex/Jellyfin + NAS storage | Compute and drive array can be upgraded separately | Second system, additional power, reliable share mounts, separate app backup | NAS storage is sound but transcoding or app compute is limiting |
| New higher-spec NAS | Solves real bay/capacity/features needs together with compute | Migration, drive compatibility, software differences and total system cost | Both storage and server platform genuinely need replacement |
A two-box design is not intrinsically faster. It only helps if the new compute hardware fixes a verified server bottleneck and the NAS-to-server connection can supply the media. Likewise, a more expensive NAS does not guarantee more supported hardware-transcoding codecs. For a specification-based shortlist of NAS media engines, see Best NAS for Plex and Jellyfin Hardware Transcoding. For the broader home-server architecture decision, see Mini PC Plus NAS vs a NAS-Only Home Server.
Calculate the Full Upgrade Cost, Not Just the Box Price
This is a diagnostic guide rather than a live price comparison. Retail prices, licences and energy tariffs vary, so the correct purchase comparison is a bill of materials based on your own verified quotes. Include the parts you already own as reused rather than pricing them twice:
| Cost category | Separate mini PC + existing NAS | Replacement NAS |
|---|---|---|
| Main hardware | Mini PC, correct RAM/SSD configuration, power adaptor | NAS enclosure with correct CPU/GPU and drive compatibility |
| Storage | Existing NAS disks; app-data SSD if needed | New drives if migration/reuse is not supported; SSD if needed |
| Networking | Ethernet cable; possibly NIC/switch if measured need | Same: NIC/switch/cables only if bottleneck proven |
| Protection and backup | UPS capacity for both devices; separate backup target | UPS and independent backup target sized to the new usable data |
| Software | Plex Pass if Plex hardware acceleration is required; other verified licences | Same software/licence considerations; vendor app support |
| Migration and support | Media database backup, share mounting, OS maintenance | Array migration/rebuild time, data copy, vendor lock-in, warranties |
| Running costs | Power draw of an additional always-on box | Difference between old and new NAS at the wall |
Illustrative electricity calculation, not a product measurement: if adding a second always-on computer raises measured wall power by 15W, the extra energy over five years is 0.015kW × 24h × 365 × 5 = 657kWh. At an example all-in electricity rate of £0.25/kWh, that is about £164.25. Substitute your own tariff and measured incremental watts. The power difference may be smaller or larger, and a replacement NAS also has a power cost. Compare like-for-like workloads, not CPU TDP labels.
Do not overlook the backup cost when buying larger drives. A new 40TB usable media pool without an appropriately sized independent backup may increase your exposure to accidental deletion, ransomware or multiple drive failures. If budget is tight, preserving the existing NAS and buying a modest compute host can free money for that more important protection.
Upgrade Decision Checklist
- Have I recorded whether the failing playback is Direct Play, remux/audio conversion or video transcode?
- Have I checked codec, audio, subtitles, HDR and playback quality on the exact failing client?
- Have I reproduced the issue with one other client and, where possible, a wired connection?
- Have I measured or observed CPU/GPU activity, NAS disk health and network throughput during the problem?
- If video conversion is necessary, is hardware acceleration supported, enabled and actually used?
- If considering more storage, do I need capacity/bays/reliability rather than assuming SSDs will cure buffering?
- Have I included licences, RAM/SSD, NIC/switch, UPS, backup and five-year electricity in the cost comparison?
- Can I solve the issue with a player change, subtitle/audio choice, schedule change or app-data SSD first?
Frequently Asked Questions
Will a faster NAS stop Plex buffering?
Only when the NAS is the demonstrated bottleneck. If the file Direct Plays and storage is healthy, look at the playback client and network first. If Plex runs on the NAS and must transcode, its media hardware may be the limitation—but that is a compute issue, not automatically a disk-speed issue.
Do I need 2.5GbE for 4K Plex?
Not automatically. A single ordinary 4K stream can fit comfortably within 1GbE. Very high-bitrate remuxes may challenge a 100Mb/s TV port or unstable Wi-Fi. 2.5GbE becomes more useful for concurrent transfers, multiple high-bitrate clients or a NAS also used for backups and workstation files.
Does Plex run faster if I move all my films to SSD?
Usually not enough to justify the cost for a modest number of sequential video streams. An SSD for Plex/Jellyfin metadata and thumbnails can make browsing more responsive; a faster video engine helps necessary transcoding. Keep large media files on healthy HDDs unless you have measured an actual storage-throughput need.
Is a mini PC better than buying a new NAS?
When your existing NAS has sufficient healthy capacity and the proven limitation is transcoding or application compute, a mini PC often lets you fix the right component without migrating storage. If you also need more drive bays, modern snapshots, better backup support or an unsupported old NAS, a new NAS may be more sensible.
Bottom Line: Fix the Bottleneck You Can Demonstrate
Direct Play plus buffering: investigate the client and network. Necessary video transcoding that cannot keep up: verify hardware acceleration and then upgrade compute if required. Slow posters and searches: examine application data and SSD placement. Low space, unhealthy disks or sustained storage latency: upgrade or repair the NAS storage and protect the data with independent backups. The most economical Plex upgrade is often the one you discover you do not need.
Related Guides
- Best NAS for Plex and Jellyfin Hardware Transcoding – a NAS media-engine buying comparison.
- Mini PC Plus NAS vs a NAS-Only Home Server – system architecture and upgrade flexibility.
- How Much RAM Does a NAS Need for Docker, Plex and VMs? – separate memory pressure from video processing.
- Best 4-Bay NAS for Docker, Plex and Family Storage – storage and application platform selection.
Official Documentation and Sources
- Plex – Direct Play and Direct Stream (playback paths and subtitle behaviour).
- Plex – Using Hardware-Accelerated Streaming (Plex Pass, transcoder settings and
(hw)verification). - Plex – Why Is My Video Stream Buffering? (network and transcoding diagnoses).
- Plex – Remote Access (upload and remote-stream settings).
- Plex – Mounting Network Resources (remote share availability).
- Jellyfin – Transcoding (playback methods and tone mapping).
- Jellyfin – Hardware Acceleration (supported backends and limitations).
- Jellyfin – Client Codec Support (client compatibility and subtitle consequences).