QTS 6 on the TVS-x71 (Linux 5.10 to 6.6 technical improvements)

Hello everyone,

I’d like to bring up an important question regarding the future of the TVS-871 and its compatibility with QTS 6.

My TVS-871 is currently running QTS 5.2.11 with QNAP’s Linux 5.10.60 kernel. Although this is an older platform, it remains a capable 8-bay NAS with an Intel x86-64 architecture, PCIe expansion capabilities and 16 GB of RAM.

What concerns me is the possibility that this hardware generation could be excluded from QTS 6 without taking into account just how much the Linux kernel has evolved between 5.10 and 6.6.

I’ve been looking through the upstream Linux kernel release notes, and there are quite a few changes that I believe would be genuinely worthwhile on this platform. I am not talking about cosmetic improvements or features that only benefit the latest processors. Several developments directly concern CPU scheduling, memory management, software RAID, ext4, block I/O and networking.

One of the most interesting changes is the introduction of EEVDF in Linux 6.6, replacing the previous CFS scheduling algorithm. This changes how CPU execution time is allocated between competing processes, with the objective of improving scheduling fairness and responsiveness, particularly for latency-sensitive tasks. On a NAS running SMB, background indexing, RAID operations, backup services and several applications simultaneously, this is certainly worth evaluating.

Memory management has also evolved considerably. Linux 6.1 introduced Multi-Gen LRU (MGLRU), an alternative page-reclaim algorithm designed to make better decisions about which memory pages should remain in RAM and which should be reclaimed. Unlike the traditional LRU approach, it uses multiple generations of page access information to improve working-set identification. This is particularly interesting for a system with 16 GB of RAM handling filesystem caching and several concurrent services. I would like QNAP to consider enabling and benchmarking this mechanism where appropriate.

There are also significant developments in the I/O infrastructure. Linux 6.6 includes an improvement to asynchronous direct I/O through io_uring, with reported throughput and latency improvements of up to 37% in certain low-queue-depth workloads. Obviously, this is not a guaranteed 37% improvement in NAS transfers, but it demonstrates the potential of the newer I/O implementation. It would be interesting to know whether QNAP’s storage services could take advantage of these developments.

What particularly caught my attention is that the Linux 6.6 release notes explicitly mention changes to the software RAID subsystem itself. For example, md/raid1 received an improvement designed to avoid lock contention caused by wake_up(). There are also changes in block-layer request handling and I/O accounting. These are not theoretical benefits for an unrelated platform: the TVS-871 uses Linux software RAID and its storage architecture is directly dependent on these components.

The filesystem layer has also received relevant improvements. Linux 6.6 includes an optimisation for ext4 delayed-allocation append writes (ext4 delalloc append write). In addition, the iomap infrastructure has received changes to improve asynchronous direct I/O by avoiding unnecessary workqueue transitions in certain paths, as well as reducing write amplification in specific block-size/page-size scenarios. These changes deserve particular attention on an eight-disk storage system handling large files and sustained write workloads.

Linux 6.1 also introduced significant improvements to BPF, including new data structures, memory allocation mechanisms and additional helpers. Linux 6.3 continued this development with TCX, an improved attachment API for traffic-control BPF programs, and further networking capabilities. These developments could provide better monitoring, filtering and network management possibilities within future QTS versions, provided QNAP chooses to integrate them.

Networking itself has also progressed. Linux 6.3 introduced IPv4 BIG TCP, extending TSO/GRO packet handling beyond the traditional 64 KB limit. Depending on the network driver, hardware offload capabilities and interface configuration, this can reduce packet-processing overhead in high-throughput environments. I would particularly like QNAP to evaluate these improvements for users with PCIe 10 GbE expansion cards and other high-speed interfaces.

Security is another important consideration. Linux 6.1 introduced KCFI (Kernel Control Flow Integrity), including support for x86-64, providing a mechanism to detect certain forms of invalid control-flow transfers. Linux 6.6 also introduced support for Intel CET shadow stacks in userspace, although this particular hardware protection requires compatible processors and would not be available on the Haswell generation. This illustrates why I would prefer a proper evaluation of each security mechanism rather than a generic statement about kernel compatibility.

There are also other developments worth considering: more configurable unbound workqueues in 6.6, improved PSI resource-pressure reporting in 6.1, additional memory-management optimisations, continued development of the device-mapper and ATA subsystems, and numerous driver updates across the intervening releases.

Of course, not every new kernel feature is relevant to the TVS-871. Some require newer processor instructions, specific hardware, particular kernel configuration options or userspace integration. My point is that there are enough directly relevant developments to justify an actual engineering assessment.

The TVS-871 uses an established Intel x86-64 platform, with the Intel B85 chipset, Intel SATA/AHCI infrastructure and Intel I210 Ethernet controllers. These rely on standard Linux subsystems and drivers such as ahci, libata and igb. I would like QNAP to investigate whether these components, together with its proprietary kernel modules and hardware-management interfaces, can be maintained on a Linux 6.6 base.

I would also like to see improvements to the native hardware monitoring capabilities, particularly the ability to expose PCIe expansion-card temperatures through QTS rather than relying on third-party scripts or manual modifications.

I fully understand that moving from 5.10 to 6.6 is not simply a matter of replacing the kernel image. QNAP would need to rebuild and validate its proprietary modules, storage management stack, fan-control mechanisms, boot environment and other platform-specific components. Existing RAID metadata, storage pools, Device Mapper volumes, ext4 filesystems, ACLs and extended attributes would also need to remain fully compatible.

However, I don’t believe these engineering requirements should automatically result in the exclusion of an otherwise functional x86-64 platform.

I would much rather see QNAP offer a dedicated experimental firmware build for the TVS-871 than see this generation permanently limited to the 5.2 branch without any technical evaluation.

I would particularly appreciate an answer from the development team on whether the TVS-X71 hardware platform has actually been assessed against Linux 6.6, and whether the main obstacles are related to hardware compatibility, proprietary QNAP kernel modules, boot firmware or simply product-support policy.

There are several years of Linux kernel development between 5.10 and 6.6. The TVS-871 should have the opportunity to benefit from these improvements rather than being excluded solely because of its age.

I hope other TVS-X71 owners will support this request and that QNAP will consider investigating the feasibility of bringing QTS 6 to this platform.

Thank you. :face_savoring_food:

Don’t hold your breath. That model was released in 2015. That’s over 11 years ago. Do you have a PC that is 11 years old? Would Microsoft or Apple support those machines today? No.

The fact that you can still open support ticket for technical help for an 11 year old machine is amazing. QNAP support is great. But everyone things that just because THEIR NAS is working and doing fine that they should be entitled to the QTS/QuTS 6 upgrade. Sorry. It doesn’t work like that.

Hardware manufacturers have to make a decision on the models to support and not support. Writing Linux drivers for old peripherals under a new kernel would be a massive undertaking. I sold processors driven by Linux before I retired and the big lift with a customer was ALWAYS getting the drivers written and working properly. That and supporting the DDR memory that was used then, etc. It’s all a HUGE lift. Updating to a new Linux kernel is far more complicated and difficult than just “updating” the kernel. People who really know Linux at a low level understand what I am saying.

If they support YOUR 11 year old model then they have to support EVERY 11 year old model. Then why not 12 year old models. Someone is going to be left out. The line has to be drawn somewhere. Sorry but that’s what it is.

I get it. It sucks. You spent xxxx dollars on it and now your investment can’t get the newest stuff. Oh well. That is just life. The cold ugly truth is that QNAP is in the business of selling new hardware not supporting models that are 11 years old.

NA9D,

I understand your point, but I believe your argument relies far too heavily on the age of the hardware and not nearly enough on its actual technical capabilities, security implications and real-world performance.

Hardware age does not automatically mean technical obsolescence.

Comparing an 11-year-old NAS to an 11-year-old consumer PC or Apple computer is not particularly relevant. A NAS is primarily a storage and network appliance. Its expected lifespan, workload and hardware requirements are fundamentally different from those of a typical desktop computer.

Enterprise storage systems and network appliances routinely remain operational for well over a decade. Their age alone does not automatically make them technically unsuitable for newer software.

My TVS-871 is equipped with an Intel Core i7-4790 processor and 16 GB of RAM. This is still a respectable hardware configuration for a NAS.

In fact, there are newer NAS models on the market today that struggle to match the processing capabilities and overall performance of this particular configuration, depending on the workload.

We should not confuse the age of a product with its actual computing capabilities. A newer entry-level NAS is not automatically more powerful than an older high-end model.

Security is becoming one of the strongest arguments for extending software support.

And this is something I believe your argument completely overlooks.

NAS security has become an increasingly serious issue over recent years. Ransomware campaigns, remotely exploitable vulnerabilities, compromised web interfaces and attacks against exposed NAS services have demonstrated how important long-term security maintenance really is.

QNAP itself has had to deal with numerous security incidents and vulnerability disclosures over the years.

In this context, keeping perfectly functional hardware restricted to an older software branch indefinitely is not necessarily a sensible long-term strategy.

Security is not just about having the latest features. It is about maintaining an operating system, kernel, network stack and associated services that can continue receiving security fixes and improvements.

A newer kernel can provide additional security mechanisms and improvements, although I fully acknowledge that kernel upgrades alone do not guarantee a secure system.

The important question is this: why should a perfectly capable x86-64 platform automatically lose the opportunity to benefit from future security improvements simply because of its release date?

I would argue that extending the security lifespan of existing hardware is actually in the interest of both customers and QNAP.

Moving to Linux 6.6 does not mean rewriting every driver from scratch.

I am not claiming that upgrading the TVS-871 to Linux 6.6 would be easy, nor am I suggesting that QNAP should simply replace the kernel and release firmware without extensive testing.

I explicitly acknowledged the engineering challenges in my original post, including proprietary kernel modules, storage management, hardware monitoring, fan control, boot compatibility and filesystem validation.

However, there is an important distinction between saying that an upgrade requires significant engineering work and claiming that it is technically impractical or impossible.

The TVS-871 is based on an Intel x86-64 Haswell platform, with an Intel B85 chipset, standard AHCI/SATA infrastructure and Intel I210 network controllers using the well-established igb driver.

These are not obscure or completely unsupported hardware components. They rely on mature Linux subsystems that have been maintained for many years.

Furthermore, Linux does not require manufacturers to rewrite every hardware driver simply because the kernel version changes. Existing upstream drivers can often be reused, although compatibility, kernel API changes and vendor-specific modifications must obviously be checked.

Your reference to DDR memory support is also somewhat misleading. Linux does not generally require an entirely new memory driver simply because a platform uses an older DDR generation. Memory-controller compatibility is primarily a hardware and firmware matter.

The actual hardware requirements of Linux 6.6 are also worth examining.

The official Linux 6.6 documentation does not establish a universal minimum hardware configuration requiring a recent-generation Intel processor or a specific amount of RAM.

The TVS-871, equipped with an Intel Core i7-4790, already meets the fundamental architectural requirements of a 64-bit x86 Linux platform.

For clarity, here are the actual hardware specifications of my system:

*Intel Core i7-4790 (Haswell, 4th generation).

  • 4 physical cores and 8 threads.
  • 3.6 GHz base frequency, up to 4.0 GHz Turbo.
  • Intel 64-bit architecture.
  • SSE4.1, SSE4.2 and AVX2 instruction sets.
  • 16 GB of RAM.
  • Intel VT-x and VT-d virtualization support.
  • PCIe 3.0 expansion capabilities.

These are not particularly restrictive specifications for running a modern Linux kernel.

Furthermore, the absence of certain newer processor features does not automatically prevent Linux 6.6 from running.

For example, Intel CET Shadow Stack functionality requires compatible processor hardware. The Haswell architecture does not provide this particular feature, but that does not make the entire kernel incompatible.

A kernel can support multiple hardware generations while selectively enabling or disabling features according to the capabilities of the underlying processor.

There is also an important distinction between kernel build requirements and runtime hardware requirements. The GCC, GNU Make and Binutils versions listed in the Linux 6.6 documentation concern the development and compilation environment, not a requirement to purchase a newer processor or install more RAM.

In other words, the question is not whether the TVS-871 meets some arbitrary modern hardware standard. The question is whether QNAP’s platform-specific components can be adapted and validated against Linux 6.6.

From an architectural perspective, I see no obvious reason to dismiss the TVS-871 before conducting that assessment.

Of course, this does not prove that QNAP’s complete firmware stack is compatible. Proprietary modules, bootloader integration, storage management and hardware-specific services still require proper engineering and validation.

Your argument about supporting every 11-year-old NAS is a false equivalence.

QNAP already maintains different firmware branches and hardware support policies across its product range. There is no technical or commercial reason why a manufacturer could not selectively evaluate individual platforms based on their architecture, compatibility and maintenance requirements.

The TVS-x71 family is not necessarily equivalent to every other product released during the same period. Each platform has its own hardware configuration, controllers, drivers and expansion capabilities.

Supporting one particular platform does not automatically mean supporting every other device of the same age.

Performance is another reason why the TVS-871 deserves consideration.
My NAS is not sitting in a museum waiting to be replaced.

It has eight drive bays, an Intel Core i7-4790, 16 GB of RAM, PCIe expansion capabilities and a storage architecture that remains perfectly relevant for many demanding workloads.

Its performance is still competitive in practical usage, and some considerably newer NAS products cannot match its capabilities in certain workloads.

Replacing such a machine simply because it is 11 years old would not necessarily represent a meaningful performance improvement.

A hardware replacement should be justified by actual technical limitations, not by an arbitrary age threshold.

I am not demanding indefinite free support. I am not asking QNAP to finance unlimited development and maintenance for every legacy device.

I am asking for something considerably more reasonable: an actual technical feasibility assessment and, if possible, an experimental firmware build for a platform whose hardware remains perfectly usable.

There is also a commercial argument worth considering.

Extending the useful software and security lifespan of existing hardware can strengthen customer confidence in QNAP.

Customers who invest in high-end eight-bay NAS systems should reasonably expect a transparent explanation of their technical limitations, rather than having their hardware dismissed solely because of its release date.

Finally, saying that QNAP is in the business of selling new hardware does not establish that the TVS-871 cannot run QTS 6.

Commercial priorities and technical feasibility are two separate questions.

If QNAP determines that the engineering effort, security validation or maintenance costs are disproportionate, that is a legitimate business decision. But it should not be confused with proof that the platform is technically incapable of running a newer kernel.

I am perfectly aware that some features introduced between Linux 5.10 and 6.6 may not be relevant or available on Haswell hardware. I have already acknowledged that in my original post.

The point is to evaluate what can actually be supported, rather than dismissing the entire platform before that assessment has even been made.

I am not asking QNAP to ignore hardware limitations. I am asking QNAP to distinguish between genuine technical limitations, security considerations and a product-support decision.

And frankly, I believe that is a much more constructive discussion than simply saying that an 11-year-old NAS belongs in the past.

Regards,

Hyperia

I don’t have time to read your book, most of which is likely AI generated. No one has the time to write two posts as long as yours.

The bottom line is that a manufacturer has to draw a line somewhere on what hardware to support. Their mission is not to support your old hardware for free. Their mission is to sell new hardware, so they can employ and pay people who will then in turn feed their families. For QNAP to support old hardware beyond a certain point is to rob their employees and rob their business. Period. You may not like that but it’s life.

I will stick to my point that writing Linux drivers for the various hardware devices are not trivial. I don’t care if they are “standard” as you say. Statements like that show how little you know about the subject. Processor architectures evolve and instruction sets get enhanced with new capabilities. Why look backward when you can look ahead.

I’m sorry you feel left out. But decisions like this will always make someone unhappy. Period. Nothing more to say on this. Feel free to put in an AI generated book though in response.

Ah, I see. When your arguments start falling apart and you can no longer defend your position, the next best option is apparently to attack the length of someone’s post, its author, and their supposed level of competence. Much easier than actually addressing the points being made, isn’t it?

As for your comment about my post supposedly being AI-generated, that’s quite revealing. Apparently, you’d rather criticize the format than engage with the substance. I couldn’t care less whether you took the time to read it or not. However, claiming to dismiss an argument while openly admitting that you haven’t even bothered to read it demonstrates a rather remarkable level of intellectual dishonesty.

I’m certainly not convinced that you’re in any position to determine what I do or do not know about this subject. Brandolini’s Law applies rather nicely here: it takes two minutes to produce a confident but uninformed statement, and twenty minutes to debunk the nonsense behind it.

And since you seem particularly concerned about QNAP’s commercial interests, let’s get one thing straight: customers also pay for software support, maintenance and security. Purchasing a high-end NAS isn’t simply about buying a box full of drives and components. Long-term software support is also part of the product’s value.

Nobody is asking QNAP to maintain every device it has ever manufactured indefinitely. I’m simply asking the company to distinguish between a genuine technical limitation and a commercial decision. Those are two entirely different things, and apparently that distinction is beyond your understanding.

Finally, telling other users that they should simply accept the abandonment of their perfectly functional hardware because it happens to suit the manufacturer’s business model is a rather peculiar interpretation of customer relations.

You’re perfectly entitled to believe that eleven-year-old hardware is no longer worth investing in. However, your personal opinion doesn’t magically become a technical fact simply because you state it with enough confidence.

So perhaps, in future, you could spend a little less time questioning other contributors’ competence and motivations, and a little more time actually addressing their arguments.

And if the subject doesn’t interest you, nobody is forcing you to participate. That’s probably the easiest way to avoid having to read something you apparently have neither the time nor the inclination to understand.

Thanks for taking the time to put together such a detailed analysis! I’ve forwarded your feedback to our product team for reference. Thanks!

I’m sitting in the same situation and would not need to upgrade my TVS871 to any newer models despite the lack of QTS6 support