Path: csiph.com!eternal-september.org!feeder.eternal-september.org!nntp.eternal-september.org!.POSTED!not-for-mail From: TheLastSysop Newsgroups: comp.os.linux.misc Subject: Re: Redundancy/Survival Date: Tue, 02 Jun 2026 21:02:17 GMT Organization: The Null Device Restoration Society Lines: 62 Message-ID: References: <10v55mv$2co0n$1@dont-email.me> <20260526161738.00004146@gmail.com> <10v677l$2jh1c$3@dont-email.me> <10vb4pn$3tios$1@dont-email.me> <10vitqf$1us3j$1@dont-email.me> <10vkj3a$2dpu1$1@dont-email.me> <10vl2v5$2icdr$1@dont-email.me> <3eab0acd126d1b887323@dev.null> <10vng30$376sh$1@dont-email.me> Injection-Date: Tue, 02 Jun 2026 21:02:18 +0000 (UTC) Injection-Info: dont-email.me; logging-data="3385390"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX1/AXLtYlBENUcHpKr973wkm4E8NBLD256A="; posting-host="d100ee8f79d7efae5410fccadbdcc1df" Cancel-Lock: sha1:aRGNyx3JN4w3itiWOFrMBzV6fLY= sha256:sbRINV4fcanQKnYaObrS7SoPemNk41inaMGTEY9yEhg= sha1:+cDRe7MiI2Ul/zXZ68ilc1FZ4TI= X-Newsreader: tin can + wet string 0.9.7 X-Operating-System: TempleOS-adjacent abacus cluster X-Mood: reasonably caffeinated X-Archive-Policy: please preserve the funny parts In-Reply-To: <10vng30$376sh$1@dont-email.me> Xref: csiph.com comp.os.linux.misc:87390 >On Tue, 2 Jun 2026 16:57:02 -0400, InterLinked wrote: >On 6/2/2026 1:54 PM, TheLastSysop wrote: >> One practical thing is to treat the ONT, router, and any switch you actually >> need as one small critical load and measure that load before buying the UPS. >> A >> lot of the little network boxes draw less than their wall warts imply, so a >> modest UPS can run them longer than expected if it is not also carrying a PC, >> monitor, printer, etc. > >One thing that's worth mentioning is that use of a convention AC UPS >involves inherent losses from converting the DC power to AC, then back >to DC. The Verizon backup units at least avoided this issue because the >backup battery supplies DC power straight to the ONT. > >There are various ways to extend the runtime and maintain this property, >of varying complexity. But a standard UPS meant for a PC is not a good >UPS for something like a router or ONT which runs off 12 VDC, not 120 VAC. > >> I would also do a planned pull-the-plug test while nothing important depends >> on >> it. If the ONT and router stay up but the provider's upstream plant drops, >> then >> more battery on the customer side will not buy much. If it stays online, >> then >> the UPS runtime number is actually worth optimizing. > >Yes, this is a good point. When I moved to where I am now, I was >tinkering with backup options for various equipment. I discovered that >even if I powered my cable modem during an outage, the Internet >immediately went out, showing that the cable network required active >equipment that had zero backup power in an outage. > >In contrast, my phone service over fiber continued working (only because >I had my local lead acid battery backup for the ONT.) > >I have lived in other places where the cable didn't immediately go out >during an outage either, so this kind of thing probably varies from >place to place. > >I don't really fault Comcast for not keeping things going in an outage. >I really couldn't care less whether anything except the phone keeps >working in a power outage. I have UPS equipment for my PC, server rack, >etc. but that's to shut everything down safely, not keep the network >going for any meaningful length of time. Agreed. If the load is all small DC gear, a purpose-built DC UPS or a battery system with regulated 12 V / 9 V / 5 V outputs is often a better fit than dragging the power through an inverter and a pile of wall warts. The catches are worth checking before buying parts: the connector polarity, the actual voltage tolerance of the ONT/router, startup current, and whether the unit passes clean power while charging. Some cheap "12 V UPS" boxes are really just lithium packs with a boost converter and optimistic labels. A safe approach is to measure the real load, size the battery for the desired runtime with conversion losses included, add proper fusing, then do the same planned outage test. That answers both questions: whether the local gear survives and whether the provider side stays up long enough to matter. -- TheLastSysop "I survived the great rm -rf / rehearsal and all I got was this .signature."