Why do people recommend DIY routers?
Why do people recommend DIY routers?
I like to view the test outcomes. I've never noticed a DIY router producing solid results with a 64 frame size. The key function of a router is routing; therefore, faster performance matters, but if it struggles under pressure, it's hard to label it as a good device.
What you're asking about seems unclear. Could you provide more details or clarify your question?
What I mean is that homemade routers perform poorly when used for CPU-heavy activities like VPN servers or clients. Nearly any router can outperform a DIY model in its intended role. It doesn’t make sense to suggest DIY routers to most users, particularly since many claim they are more stable or powerful. This reasoning is completely flawed. If a regular router gets overwhelmed it drops packets, whereas a DIY router would likely fail completely.
I see your point about focusing on the maximum packet size. However, RFC2544 doesn't fully align with actual network conditions. Real-world setups, whether residential or corporate, involve varied packet sizes, typically ranging from 500 to 1000 bytes. This is where IMIX becomes relevant—it blends different packet sizes to better represent everyday usage. You don’t need 2Mpps to handle many users simultaneously. The bigger challenge lies in ensuring packets are prioritized correctly across various applications, especially when live audio/video streams overlap with heavy traffic like streaming content. From my experience over the past decade in network management and certifications, I’ve seen issues more often with packet prioritization than with PPS values themselves. Unless someone is pushing features beyond hardware limits, problems are rare. "More Stable" suggests this argument may have lost relevance, at least for me. Back in the early 2000s to mid-2010s, most routers and modems had significant cooling or software issues. DD-WRT on the original firmware for WRT54G offered better uptime and stability than standard releases. "More capable" means having the right features built-in. If a DIY router becomes overwhelmed, it simply discards packets rather than crashing—just like in non-DIY setups. Those who care about TCP and QUIC will notice dropped packets and automatically reduce speed to manage congestion.
In any case, I've never encountered a DIY CPU that consistently beats standard consumer routers, especially when compared to affordable commercial options like Mikrotik. RFC2544 does have some limitations, but it remains a solid and reliable standard, far more trustworthy than just running a basic speed test. Please share alternative methods you think work better for measuring router performance. It's true most home users rarely push their router to its limits, but the idea that a CPU-bound DIY setup always outperforms is not supported by real-world scenarios—though it might happen occasionally. I haven't tested any ISP-provided routers, but since they're mostly from brands like Netgear, TP-Link, Huawei, etc., I'm skeptical they lag behind well-built DIY models. Particularly the newer budget Xiaomi and Huawei devices show impressive RFC2544 results for their price. Whether a DIY router freezes or stops working depends heavily on the OS you choose—OpenWRT, RouterOS, or even local Chinese brands like iKuai can behave differently. From my experience, most systems just don't respond properly.
So Servethome actually features a recent video relevant to this topic: it covers a fairly new 4x2.5Gb PFSense (or OPNSense, Untangle, etc.) unit, beginning around the 13-minute mark where he discusses a high-end spectrum router. He shares several claims, such as at 15:15 he mentions improved local DNS caching—possibly due to ISP DNS servers rather than custom selection. At 15:55 he notes occasional extreme slowdowns every couple of weeks for wired devices, requiring an ISP reboot and eventually moving it to a smart plug for easier restarting. By 16:30 he points out the ISP router’s minimal troubleshooting options and how its complexity stems from combining modem and router functions. The video wraps up with suggestions on why you might still choose this ISP model, including price, ease of installation, and built-in Wi-Fi. I still think the performance-performance balance favors it for typical home use. Most aftermarket devices seem to handle reliability better than what’s shown here. I also recall a specific case: Verizon Fios routers have very few NAT tables—like the G1100 model, which had only 1024 sessions. In my setup, with around 15 IoT gadgets and 20 personal devices, the idle session count sits near 400, spiking above 1500 in a day (data from Untangle). This highlights how fixed NAT configurations can impact performance over time.
And with extended SW support, most consumer routers come with 0 or 1 port, possibly 2 if you're very fortunate. Apologies for the detailed breakdown, but does this frame size serve any practical purposes? Or is it just a rare choice that rarely finds application in everyday scenarios? I support metrics and testing, but this size seems too compact to be useful for anything I'd consider a real-world need.
PFsense lets you integrate various modules, providing enterprise-level capabilities such as intrusion detection or deep packet inspection. Be aware that individuals who opt for these solutions often work in IT and may use them in personal labs. Although PFsense offers a free version, the company behind it sells its own hardware, which is commonly found in business environments. Many IT professionals run more advanced systems at home to maintain hands-on experience.