Jump to content

colir

Member
  • Posts

    4
  • Joined

  • Last visited

Awards

This user doesn't have any awards

Profile Information

  • Location
    London, GB
  • Interests
    Networks
  • Occupation
    Telecoms
  1. On that topic - it could be interesting for Luke / @jakkuh_t / @Emily Young to implement IPv6 in the LTT & Labs office network (and maybe also Linus' home network?). Seeing LTT videos covering the firewalls, IP allocation etc. I don't recall anything about IPv6 — almost no mentions of IPv6 on the forum either. The world is very clearly moving towards IPv6. Even this message is being written using IPv6 (since CloudFlare in front of the LTT forum is using this). We can only expect more resources and networks utilizing IPv6 since there are fewer and fewer IPv4 IP addresses available as the Internet is growing... AWS also just announced they will be surcharging all their customers $3.65 US per IP per month for the use of IPv4s (citing the rising cost of procuring IPs on the secondary market as almost none are left on the primary) and seeing this introduced by the largest cloud provider in the world, we can expect even quicker adoption of IPv6 soon. Additionally, many large ISPs (including Starlink) also already implement CG-NAT (Carrier-Grade NAT) for their IPv4 traffic meaning the customers on their home routers just see private IPv4s on the WAN side (usually from 100.64.0.0/10 block). Compared to IPv6 traffic, all traffic from such customers to the Internet using IPv4 has to go through additional (NAT) devices before it leaves the network of an ISP (which also means the traffic is NAT-ed twice as the first NAT is within home/office network). Because of this, the latency from the same client using the same ISP to the same server on the Internet will be higher when using IPv4 than IPv6 (explaining Google's results) - usually a couple of ms. A lot of ISPs nowadays tend to run IPv6 + IPv4 at the same time, it's a setup known as dual-stack (with the IPv4 part being either a public IPv4 or a private IPv4 behind CG-NAT/customer router's NAT). It looks like LTT's ISP - iTel - already supports IPv6 (https://bgp.he.net/AS16696#_prefixes6) so it should be possible for LTT to easily obtain some toasty /48 IPv6 block and implement it internally. Additional points for implementing DNS64 and NAT64 (at least as a test) - allowing to shut down IPv4 on the client side completely and yet still have access to IPv4 resources (multiple large mobile/cell carriers around the world already started doing that on a large scale! Eg. Telstra in Australia - https://en.wikipedia.org/wiki/IPv6_transition_mechanism#464XLAT). This topic could also cover how to secure access to IPv6 devices from the outside (as technically in IPv6 every single device receives 2 public IPv6 IPs). If you would like to chat more about this, feel free to send a DM PS. Fun fact - IPv6 was invented before IPv4 NAT was invented More info here: https://youtu.be/GLrfqtf4txw
  2. Hey @jakkuh_t From my experience working for an ISP, the browser version of speedtest.net is throttled a bit (ads on the screen etc.) and you are likely to get lower results, especially on slower machines. An easy way to overcome that is to install their lightweight standalone Windows or Mac app: https://www.speedtest.net/apps/windows https://www.speedtest.net/apps/mac You also won't need to cover your IP address with a finger You can also use their Linux CLI app if you want to test your bandwidth speed directly on your firewall - https://www.speedtest.net/apps/cli Wouldn't be surprised if you manage to get a ~10 Gbps result using these apps. BTW, FS.com also sells these inexpensive fibre optic connector cleaners - you probably should use those particularly for outdoor runs - https://www.fs.com/products/39721.html PS. Whilst it's a good practice to use them, I don't think you necessarily need attenuators on these 2km links, these kinds of transceivers (even 10km transceivers) are widely used in data centres without attenuators for years with no issues.
  3. Hi Theodor, RB4011 is a great piece of hardware - so it's a good start. Capacity-wise it might be a bit overkill (unless you really require over 1G capacity), if you need something under 500M and you want to save a few bucks, you might be good with slightly older RB2011. 1. It depends on technology. If your ISP is using just regular GPON or EPON (GPON is the most likely option), this might be a bit tricky - in xPON world, the end device which is being delivered by the ISP if pre-authorized to get access to their network. If you ISP is willing to give you an SFP transceiver or authorize your private xPON SPF transceiver, that might be an option. If the latter option - then you can use something like this - https://mikrotik.com/product/SFPONU. If the technology is not GPON/EPON - it's most likely a regular P2P connection using WDM. 2. SC is just a connector, there're SFP transceivers for SC. If technology is GPON/EPON - see my answer above. If the technology is P2P - you can get an SFP WDM (WDM because you mentioned 1 fibre) transceiver with a SFP connector. Then there's the type of the connector (APC or UPC), type of fibre (single mode or multi mode) which is also important before you choose a transceiver. If you feel lost with all these acronyms, just post photos of your network kit and we'll figure something out. Note: I've referred to SFP many times even though RB4011 has SFP+ port as SFP+ is compatible backwards and unless you need a 10G transceiver, you'll be good with 1G(-ish) SFP transceiver.
  4. In this video Linus has mentioned that they'll get 10G to Vancouver Internet Exchange Point - I guess it's VANIX (unless there's other I'm not aware of), is that right? The thing is - Linus said that Amazon's network is also available through this IXP which I don't think it is correct - they're not listed on https://vanix.ca/participants/. Also, the other thing which bothers me slightly is - are they gonna run their own BGP - get their own ASN and join as a regular member of VANIX or "proxy" it through Telus?
×