Upstream bandwidth in the time of COVID-19

Article By : Brian Dipert

Pandemic-induced desires for faster upstream bandwidth were the fundamental basis for this engineer’s service upgrade odyssey.

I’ll begin this post, as I do sometimes, with a flashback. Long-time readers may vaguely recall that back in mid-2007, while still living in Sacramento, CA, I briefly test-drove SureWest’s 50 Mbps fiber-based symmetrical broadband service, part of a triple-play bundle that also included VoIP and IPTV. The “infriggincredible” reaction I penned in response to my initial post-installation testing results still makes me smile; keep in mind that I wrote it more than a decade ago, and that my benchmark for comparison was my then-ADSL service (1.2 Mbps down, 300 kbps up). And the bandwidth boost I achieved (40+ Mbps down, and as high as 63+ Mbps up) is nothing to sniff at, even today, though the cost-to-consumer of that bandwidth has thankfully dropped precipitously.

Fast-forward 13+ years, and here we all are, still stuck in our homes in the midst of a lingering, raging pandemic. In my family’s case, although we don’t have any kiddos constantly slurping from broadband, both my wife and I work from home (and have since long before COVID-19 hit U.S. shores, actually; what’s changed is that pretty much everyone else is now home-based, too). Lucky her; she’s been able to dodge (at least so far) the “Zoom meeting” bullet. But the Verizon femtocell on which she’s heavily dependent taps into our common broadband pipe, of course, and is adversely affected by other heavy traffic on that same pipe. And I’m increasingly the “heavy traffic use” culprit (unluckily unable to avoid frequent sessions on Zoom, along with Webex, Skype, Microsoft Teams, Google Meet and the like). Not to mention the fact that I’ll be slinging around way more HD video and other heavy data payloads in the weeks and months to come.

Upstream bandwidth limitations were the key hindrance here, to be clear; downstream was still fine for our purposes. But something had to give; my wife’s phone calls kept getting glitchy every time I hit our broadband service hard via webcam and/or HD screen sharing usage, and my uploads were also taking way too long to complete. It was time to do something I’d long dreaded: dealing with Comcast’s customer service (translation: sales) folks to negotiate an upgrade. Some background: in my initial Colorado residence location, I’d had Comcast’s “Blast Plus” service tier, initially 25 Mbps downstream (I don’t recall what the initially-claimed “up to” upstream speed was). In 2013, Comcast boosted the “Blast Plus” speed at no extra cost, as long as I had a DOCSIS 3.0-compliant modem, to 50 Mbps down/10 Mbps up. And in late 2014, they doubled the peak advertised download speed again.

When I moved the service to our current residence in conjunction with getting married in late 2014, Comcast migrated us to the company’s “Performance Pro” tier, which supposedly supports up-to 200 Mbps downstream speeds, along with 5 Mbps upstream transfer rates. And for a while, in fact, I was getting that exact download bandwidth (a few Mbps more than that, actually). That said, a few months ago I noticed that my downstream bandwidth had dropped to <100 Mbps; I suspected that Comcast had inadvertently (?) repartitioned my modem for the “Performance Plus” tier, but I hadn’t yet gotten around to doing anything about it. Did I mention how much I disliked dealing with the company’s customer service-i.e.-sales team, who can’t seem to let go of the idea of trying to upsell you on more expensive TV channel packages, settop boxes (STBs) instead of my CableCARD, bundles that include VoIP and/or security services, and a rental-vs-paid-for modem-plus-router?

screenshot of internet speed report showing 5 Mbps upstream

Well, now was the time to finally bite the bullet. With previous frustrating customer service experiences as my guide, I told a bit of a “white lie:” from the initial telephone system menu, I selected the option that indicated I wanted to “cancel my service.” Within a minute or so, I transitioned from “on-hold” to a live conversation with what turned out to be a pleasant and helpful customer service representative (see how “dangerous” assumptions can be sometimes?). And less than 15 minutes later, my new service package was activated and ready to go. The two relevant broadband upgrade options to relieve my upstream constraints were Extreme Pro (600 mbps down/15 Mbps up) and, for only $10 more per month, Gigabit (1 Gbps down/35 Mbps up, which Comcast refers to as Extreme Pro+). You already guessed which option I picked, didn’t you?

The only minor glitch I experienced during the call was that the customer service rep initially set me up with an STB instead of the CableCARD that I’d explicitly mentioned several times up to that point, but I digress. I noticed the discrepancy in the TOS confirmation page whose URL she’d texted me, and she resolved the error in short order. The best part: even after the 24-month $10-off promotion period ends, I’ll be paying notably less than I was before, for more. And the “extras” I’m now getting extend beyond just the higher broadband speeds, but I’ll save the “rest of the story” until the end.

Even prior to getting back home (yes, I did all this while out walking the dogs) and swapping out my tried-and-true refurbished Linksys DPC3008 cable modem that had been in my possession since late-2012 (and was still working like a charm, but has now been relegated to the future-teardown queue), my downstream speed immediately jumped back up to the >200 Mbps range, although upstream speeds didn’t budge:

screenshot of internet speed report showing 200 Mbps downstream

The upstream bandwidth non-movement was my first indication that something was amiss. The DPC3008 supports 8 bonded downstream channels and 4 bonded upstream channels (“8×4” in DOCSIS lingo), which theoretically should translate to 108 Mbps peak upstream speeds. As noted in that Quora post: “These channels are typically shared by multiple subscribers for cost reasons and limited by things like cable quality and noise. The maximum speeds are theoretically possible, but would not be cost effective to deliver to individual users.” Still, <6 Mbps upstream represents a significant undershoot to this theoretical peak.

I’d originally planned on swapping out the DPC3008 for a NETGEAR CM700 (32×8 DOCSIS 3.0) that I’d bought on sale refurbished for $64.99 at Newegg, but I’m glad in retrospect that I did a bit of research prior to proceeding with the transition. A quick aside, by the way: as a general rule, it’s a bit risky to pick up a refurbished (or more generally used off Ebay, etc.) cable modem, since if the previous owner’s cable service provider hasn’t de-partitioned it, you won’t be able to re-activate it yourself. That said, since the brand-new MSRP on the CM700 is $109.99, and since the product description noted that it was a “NETGEAR Certified Refurbished product,” I figured it was worth the risk.

As it turns out, Comcast caps upstream speeds for all DOCSIS 3.0 modems on its Gigabit plans to 6 Mbps, regardless of what channel-bonded speed the modem is actually capable of supporting; a DOCSIS 3.1 modem is required. So instead, I went with another product in my just-in-case stash, a NETGEAR CM1000 (MSRP $199.99, 32×8 DOCSIS 3.1) that I’d picked up on sale at Costco in early 2019 for $119.99.

photo of the NETGEAR CM1000 modem

It’s admittedly a bit overkill for my current needs, since it includes dual GbE WAN connections (therefore the “multi-Gig” claims on the product page). But considering that it was notably less expensive at the time than the single-WAN CM1000, I took the bait. And in retrospect, I’m glad I did. I gave the Comcast customer support representative the new cable modem’s MAC address for partitioning purposes, and she proactively prompted me to confirm the old modem’s MAC address so she could de-partition it at the same time. I powered down the old modem, plugged in and powered up the new one in its stead, waited a few minutes (which included an auto-reboot in response to Comcast’s partitioning command to it) and here’s what I saw:

screenshot of internet speed report showing 600 Mbps downstream

Bingo. My OnHub router automatically checks and logs network performance every day or so, and for the first few days, I also periodically did manual analyses. Downstream speeds have occasionally dipped into the 400s as well as risen further into the 600s:

screenshot of internet speed report showing 400 Mbps downstream

screenshot of internet speed report showing 623 Mbps downstream

Note, however, that the 39 Mbps upstream speed (which again was my primary upgrade motivation, and the more important of the two parameters) has remained steady at 39 Mbps throughout. Note, too, that these speeds are measured at the router itself; LAN congestion and other factors may constrain the actual speed that any particular network client will experience (that said, from the get-go I built my home LAN to be fully GbE capable from both switch and cabling standpoints).

Before wrapping up, let’s revisit my earlier “extras” teaser. For one thing, although I’m still paying $15/month extra for HBO, Showtime (which I didn’t have before) is now bundled for free. Also now bundled is Comcast’s HD DVR service. The “DVR” bit is of no current use to me, since I don’t use Xfinity STBs; yes, I’m still stubbornly limping along with my Windows Media Server- and multiple Xbox 360-based setup, more than five years after prematurely bidding it goodbye. But the “HD” part is quite nice; with it, I can alternatively access Comcast content (that I’m subscribed to, of course) on my Rokus in conjunction with the Xfinity Stream Beta app. And that’s particularly useful since Windows Media Center can no longer stream now-MPEG4-encrypted HBO content in HD, since HBO has retired its HBO Go service, and since Roku and HBO have yet to come to a carriage agreement for the HBO Max successor.

1,600+ words in, I’m sticking a fork in this one, it’s done. Over to you, readers, for your thoughts in the comments!

This article was originally published on EDN.

Brian Dipert is Editor-in-Chief of the Edge AI and Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.

Related articles:

Leave a comment