Workplace reviews and discussion: QNAP_威聯通科技股份有限公司

I just lost two thousand-character drafts because of the terrible interface design, and this is my third time typing it. It really does embody “think twice before you act”—thanks, Good Job Taiwan (go to hell).

The following is the opinion of a former employee. The last time I had insight into the company’s situation was last year, so please take it for reference only.

==

QNAP uses a BU-based structure: under one large company are various R&D teams, each led by its own BU head. Everything from benefits to working hours is managed by each BU independently. That’s why when many people talk about QNAP’s benefits, it can feel like they’re not talking about the same company.

There is also a PM group that coordinates among the various BUs.

(And MIS is another independent unit.)

The advantage of the BU system is that each BU has its own technical talent across various fields—apps, frontend, design, backend, and so on—and each BU is responsible for its own scope, such as different NAS apps.

The BU system is task- and product-oriented, so the quality of individual products is relatively good. Also, because collaboration is necessary, the basic integration interfaces aren’t too bad. The downside is that integration between NAS apps isn’t very good and requires project-based coordination.

However, NAS apps don’t really need that much integration with one another. As long as they connect at the storage and Web API layers, it works, so this architecture is quite suitable and reasonably elegant.

The thing people most often criticize internally is that the design elements (CI) aren’t very consistent, but honestly, I personally don’t think this is a major problem.

Another issue is that when a BU is poorly structured, the things it makes tend to be especially terrible in its weak areas. For example, when QFile uses the web to share files with others, the interface is beyond awful. Even the download button can behave strangely.

For example, if you set up a folder link and enable file uploads, opening the link does not immediately show an upload button. You first have to open a folder, and only then does the upload button appear. That is completely wrong for the intended use case.

For example, on a file-sharing download link, clicking the “file name” does not download the file. You have to specifically click the “Action” link. I especially invite you to open a URL on my NAS and experience this bizarre behavior for yourself:

http://tonyq.myqnapcloud.com/share.cgi?ssid=0shLWLZ

I filed a bug report about this internally back in 2014. I assume it has already been classified as a feature designed to entertain people, rather than a bug.

Even so, QNAP still does very well in the BUs where it is strong.

As I understand it, S company mainly uses a division-of-labor model organized by specialties: one department handles frontend, another handles design, another handles backend, and so on. This approach has advantages and disadvantages, but for NAS development in the form of apps, I still prefer QNAP’s BU system. In my view, QNAP’s hardware and driver foundations are also better than S company’s. As a hardcore engineering nerd, I prefer hardware with refined components and stable, reliable performance over products whose interfaces may seem considerate but whose foundations are concerning.

I actually think QNAP has continued to benefit from the excellent driver architecture it established early on. Another factor is that some products were simply done right—for example, myQNAPcloud is a great design. (But please don’t use the HTTPS add-on as a money-making tool. Why be so greedy? Charging a reasonable fee to cover costs would be fine, but why charge that much?)

The team led by someone nicknamed “Doctor” previously used a very solid approach to driver development. So although QTS has undergone many revisions over the years and occasionally experienced minor turbulence, at least it has generally avoided major problems. I think that’s pretty good.

The trade-off is that, because they weren’t familiar with the web side, they didn’t establish a good configuration mechanism. As a result, the built-in virtual host file constantly has problems with different parties overwriting one another’s changes.

As for the advantage of the BU system mentioned earlier: at the end of 2014, Teddy began actively intervening in QNAP’s operations. He gradually sidelined Ming-Chih and removed the people around him, leaving the entire BU organization relatively inexperienced, and then imposed excessive micromanagement. (Someone commented that “the specs were written by Teddy”—that’s what they meant.)

Teddy was continually dissatisfied with the status quo, but he failed to recognize the fundamental problem. His approach left the BUs constantly scrambling and uncertain about their futures. The advantages of the BU system were severely damaged, while its disadvantages continued to worsen.

===

I believe the fundamental problem is that NAS apps are a kind of investment in the future. Users’ actual NAS needs are really just storing files, having network file sharing, accessing the NAS from their phones, backing up phone data, and at most downloading BT or other internet data. That’s about it.

I tried NAS virtualization several times myself. I even asked a colleague and added more RAM to the NAS, but the results were only so-so.

There are also many considerations involved in using MySQL or httpd on a NAS. Basically, at the time we proposed moving toward an engineer-oriented version, with some things to be gradually separated out.

Later, after virtualization and Bitnami were combined, at least some applications could be used directly, making it feel more like a coherent solution. These were things the various NAS app BUs worked on independently.

But this brings us back to the issue that the number of people downloading these apps is vastly smaller than the number of people who actually buy NAS devices.

In other words, most users may not need these functions at all. But this is actually an investment: continuously developing these apps, connecting them to users’ real-world scenarios, identifying which ones are more useful, and eliminating the unnecessary ones.

So can they simply “not develop” new features?

No. Although consumers may never use them, they still like comparing checklists to see which company has this feature and which company has that feature—“even if they never use them!”

Under these circumstances, the constant reshuffling of the BUs is completely understandable.

But the most frightening thing is that NAS devices will sell regardless! That’s good for the company, but for R&D, many departments may have no idea what users are actually willing to pay for. XD

====

The problem with Teddy’s leadership was that he believed the BUs weren’t doing a good job and users didn’t want to use their products because the BUs had made things users themselves didn’t want. But based on the times I observed Teddy’s decisions, he simply wasn’t a normal NAS user. The scenarios he imagined were far too disconnected from the front line.

He then became dissatisfied with everything the existing teams did, and several power struggles played out along the way.

Teddy / Richard / Ming-Chih / YT—in any case, no matter what QNAP does, NAS demand will still allow them to make money for quite some time.

Is this company worth joining?

It depends on whether the BU head you’ll work for is favored and what their relationship with Teddy is like. It also depends on how much your work affects them and whether you provide direct assistance.

This company really is an interesting place. I had not graduated from college when I joined. After interviewing me, my manager (the BU head) offered me an exceptional salary—60K or 65K, I can’t remember. I also had a special interview with Ming-Chih, and once he nodded, I had no trouble getting in.

As long as the BU head is reasonable, many things—including benefits—are actually easy to negotiate.

It’s a place somewhat like going to war with an army: there are various regional warlords. For people with strong combat ability, especially those capable of fighting independently, it can be an interesting battlefield. For people who only want to follow the organization and do their work, it can be a somewhat difficult place.

I think the problem QNAP originally set out to solve was genuinely compelling. If possible, I hope to see it continue to develop and improve, and management really needs to recognize the issues with the product and the company’s positioning.

And they need to acknowledge one thing: you can’t buy capable engineers who are genuinely motivated to push forward simply by throwing money at them. You also need to provide room and authority.

This is partly a record of, and a bit of nostalgia for, my former employer. I wasn’t there very long, so if I got anything wrong, please correct me.

And the interface for posting comments on the Good Job Taiwan website is absolutely terrible. Please fix it. I typed this post three times, and on the third attempt I finally had to write it in Sublime before it would submit successfully. (Angry)