Nick Buraglio

A new House proposal would divert BEAD funds into LEOS and FWA hands for equipment and to "buy down" the service cost for the first year. Representative Taylor of Ohio says this will be a stopgap measure while fiber networks are built. But there are two problems with this approach. First, if these technologies are providing 100x20 service to an area, why would anyone build fiber? This will result in these areas NEVER getting fiber and falling behind in the future. Plus, would these technologies commit to serving 100% of the locations in an area (a BEAD requirement)? Second, if a majority of locations take the FWA or LEOS service, they may be unwilling to change over to fiber once it is available. This means the business case for fiber in Rural areas gets worse, not better. A stopgap, in this case, means a future of Internet haves and have nots in Rural America, which is what we have today. Let's focus on getting fiber out to everyone where it makes economic sense and not force folks to take a lesser service just because it is cheap and fast. https://lnkd.in/gez4kjbe

13

9 Comments

At #NANOG95, Rene Santiago from Ezee Fiber shared their network design for an active/active CGNAT aka Large Scale NAT appliance pair. I use the term CGNAT here, but Rene suggests that LSN and CGNAT are interchangeable. For brevity, let’s assume you know the reasons you need to deploy CGNAT and are aware that some of the historical complaints of CGNAT’s impact to subscribers have been addressed. Let’s move to the core problem Rene addresses. How do you get traffic from the subscriber to the CGNAT appliance and back? There are many industry approaches Rene discussed. You can use a dedicated L2 network, PBR hop by hop (if you’re brave, as it’s labor intensive), multi-VRF routing (what Rene sees frequently deployed), PBR at the border only where the cluster sits (easy to implement for small operators, hard if you are large and have geographic spread), and pseudowire headend with PBR. While Rene was aware of all of these existing solutions, his requirements were for simplicity, no VRFs, scalability, minimal touchpoints, resiliency, and a centralized solution. Why centralized? He needed a strategic location where, for an active/active deployment, he could have all the NAT states sync between the CGNAT devices. This is what Rene came up with. As quickly as he was moving, I couldn’t capture all the details of which there were many. Here’s what I got! Traffic flows from the subscriber to the Internet… 1. PBR at the edge router. Look at the source IP, and point it to the anycast loopback the routers are advertising. Add the appropriate SR label. 2. Route it to the core using SR and pop the label (PHP). 3. PBR at the core routers looking at the source IP so it goes towards the appliances and not the Internet. 4. NAT and native routing. Traffic is NATted by the CGNAT appliance and then routed to the Internet. Return traffic… 1. Route to the appliance where it’s unNATted. 2. Send it back to the edge router via SR. Then natively forward it back to the subscriber. Rene described the active/active CGNAT appliance architecture… 1. Split a block in half, where one loopback is anycast for the first half, and a second loopback for the second half. Each appliance is standby for the other appliance’s active half. 2. Balance the load and have failover via IS-IS metrics. Add a weight of 1000 to the opposite sides. 3. Use PBR to steer traffic to the CGNAT appliance that owns a given half. Rene then talked through the routing that happens during failure scenarios. For example… 1. What happens when you lose an uplink to one of the core routers? 2. What happens when you lose one of the active/active CGNAT appliances? 3. What happens when you lose the state mirroring/VRRP link (split-brain) between the CGNAT appliances? 4. What happens when you force a single appliance to be active for both halves? (You have to update the IS-IS metrics so that routing is correct in Rene’s architecture.)

12

The most dangerous gap in America isn't geographic! It's digital. The digital divide isn't about distance from a city.  It's about distance from opportunity. A kid in rural Montana with fiber internet has more career options than a kid in downtown Detroit with unreliable connectivity. A small business owner in a connected farming town can compete globally. A shop owner in an unconnected suburb gets left behind. This isn't about zip codes. It's about access to: 🌳 Educational resources that don't require driving to a library  🌳 Job markets that extend beyond your physical location 🌳 Healthcare that reaches you through a screen when you need it  🌳 Economic opportunities that aren't limited by local businesses  🌳 The freedom to live where you want without sacrificing your career When we call it "rural internet," we're thinking too small. This is about creating two different Americas: One where opportunity flows freely, and one where it gets stuck at the last mile of infrastructure. Geography used to determine your life's possibilities. Connectivity should make geography irrelevant. That's not a rural problem or an urban problem.  That's an American problem. And it's solvable. 🇺🇸

30

18 Comments

The FCC this morning adopted a rule dropping the requirement for a “certified professional engineer” to certify twice-yearly data submission to the Broadband Data Collection (BDC) process. ISPs, especially rural broadband providers, found it difficult to find professional engineers for the task and the Commission has regularly waived the rule since BDC’s implementation in 2022. The replacement rule allowing a “qualified engineer“ to certify is “a more flexible approach,” said newly sworn-in Commissioner Olivia Trusty attending her first FCC meeting. https://lnkd.in/eQ-8TDGZ

24

1 Comment