up_the_irons: yeah, in the case of my pair of chassis, it's reserving for two additional supervisors, rather than just the standard single redundant sup so...500W for one Sup2T, then reserving ~1000W for two more I think it's a bug also, yup, MX80 = 372W also does N+1 PSUs rather than quasi-redundant power supplies like in cisco gear - they load share, and you can power the whole device off 3 PSUs, but there's also a fourth one in the chassis that serves as the +1 well, i guess not less than 350W (too early for me to be reading) but /close to/ 350W :) Anyone know of a website that specializes in the storage, collection, archiving, search and retreival of scanned images and documents? tagging functionality as well as folders / labels would also be handy m0unds: wow that's pretty cool of the MX80 spec-wise, they're pretty impressive too i'd like to mess with more j gear, but every site i've worked with is 100% cisco haha, only $40k for an MX480 "only" jeez, for $40k, you can also get an EX8216 (16-slot EX8200 switch chassis) w/redundant PSUs, 3x 10GbE SFP+ line cards, 2x 48-port SFP line card and 2x 48 Port 10/100/1000 line cards yeah the pricing is always crazy, which is why i never have the opportunity to get a juniper just need to find a tipped over delivery truck full of j stuff wont be hard it's not like many people will be grabbing at stuff compared to an overturned alcohol truck or something you could probably take the whole truck hahaha and a couple guys to help you move stuff - those switch chassis' are heavy heavy heavy LOL i actually found a 6506 chassis in a dumpster once. i grabbed it. ;) Never had the courage to use it cuz wtf, it could be garbage for real. had a big dent too, but could probably be hammered out. hah do.. you still have it? m0unds: the J stuff isn't good for the modern internet mercutio: and i know tons of folks who work on huge modern networks who would disagree even MX120 sucks with full convergence max120 is tiny mx120 juniper don't have SMP support yet adn use small cpus cisco use small cpus too, but have more efficient bgp implementation they use appropriately specced cpus in their routing engines and other stuff ask the folks who bought into nexus and ios xr how much fun that's been for them, haha well route tables are going to get bigger, and if you want to do things like rpki it uses more cpu i dunno i use bsd :) and how cisco's own engineering folks realized they screwed up and had to reengineer an old product (6500) and release that because their flagship line is pretty terrible i don't think one datapoint makes something better or worse than something else well bgp convergence times are too long on the current internet and with things like ddos attacks taking out transit links at times, or people just screwing up fibre connects, bgp convergence time is noticable. also, junos has supported SMP since 8.5, if their release notes are trustworthy curious mx120 has two cpus only one used from what i read SMP support in the kernel was added in 8.5 via an upgrade to the kernel and control plane (freebsd) maybe i thinking of mx80 i look eat list and i can't see 120 last time i see anyone mentioning of it not supporting SMP was in 2011 http://web.archiveorange.com/archive/v/M4zXgZAHIfCG3dgyuFF1 i'd check if i had an MX80 in front of me yeh i think it's mx80 i was thinking of which still arne't cheap the bgp convergence issue was due to rpd, and according to jtac that was fixed in 11.4R7-S1 when was that? some weird blocking condition looking i think it's 2 or 3 minutes when not foobared and 15minutes plus when foobared but it means people are hesitant to use full route tables on them haha, they still haven't fixed my juniper acct march of 2013 they also don't have reundant re larger chassis' do yeh can accommodate it anyway then you have to pay for line cards yep and the cheap line cards don't do aqm and the aqm line cards cost more but i suppose networks is something poeople will invest in yea, biggest capex here is networking equipment i do think they're overpriced in general and maybe this sdn stuff will improve matters i think a lot of the dedicated networking gear is way expensive in some cases i think it's justified, and in others i think it's absurd cos it'd be nicer to have a pc or such control the routing and just tell the switch where to send stuff i want to see more interconnects everywhere myself like between providers local peering etc which means terminating connections for local users close too which means needing a router or something close too, whiich means people are likely to use smaller routers http://support.arpnetworks.com/kb/main/is-there-a-firewall-filter-rate-limit-or-similar-device-applied-to-my-traffic is great "SSH rate-limit inbound on port 22 (VPS only)" now i don't need to run fail2ban on my vps's awesome that it's documented too Yep. is it still good practice to run pf for basics such as filtering all port 22 access that doesn't come from my isp? stupid question: is using something like this to cool the a server room/closet considered to be a sane way to do things: http://www.tripplite.com/sku/SRCOOL12K/ or this: http://www.schneider-electric.com/solutions/ww/en/med/4664183/application/pdf/487-srio-7h8gkg_r0_en_src.pdf goodwill: we use similar units as backup cooling (in the event that our chiller plant has issues) goodwill: same as m0unds we have 3 of them. but they are for emergency/backup only. if you don't setup a permanent drain line, there is a "bucket" that catches the water. If yuo don't drain it often enough, it has a switch that kills the unit so you don't flood the server room. In florida, that bucket fills in under 8 hours. luckily, i'm in the desert and there's little to no condensate to drain, hahaha m0unds: jpalmer : a lot of those units don't have the drain buckets anymore, they just evaporate it somehow huh. we've got a big-ass backup AC for our server room - it has a garden hose-sized drain on it. i think it's like a 3 ton unit or something ridiculous like that. takes two people to move it, despite it being on wheels.