Tuesday, June 8, 2021

Network Access Control for wifi, segregating employee, byod, guest, iot, etc.

We need to implement NAC on our wifi networks. Still using a shared key and changing it every 3 - 12 months or after a lot of people quit. But would love to get NAC working before COVID ends and people return to the office.

We bought Ruckus Cloudpath over a year ago and never got it working. Tried working with support and our vendor but can't get enough instructions to have it actually function. And then a few days ago I tried to upgrade it and the VM died. It looks like it's booted but can't login or access management page, etc. Tried restoring snapshots and backups, same prob.

So if I have to rebuild might as well start new.

We have two offices. Pre-covid populations had about 40 people in one and 80 - 120 in the other.

We want to segregate company-owned laptops & devices from employee phones/BYOD, guest access, and IOT. All APs are Ruckus.

What are your favorite, functional NACs?



Networking pros and cons

What are the pros and cons of making a network completely uniform in the choice of hardware and software?



P4 Barefoot Tofino

Hello there,

I want to write P4 code for Intels Barefoot Tofino switches but don't really know where to start learning the differences, on which page I can learn about it best, respectively. Hope you guys can help me. Thanks!



Super Simple Maps

Hello folks,

Many (many) moons ago, I used to have a really useful bit of software called WS-Watch. I’m pretty sure it got bought up by Ipswich to become WhatsUp Gold, but might be wrong.

I was a really simple bit of software for drawing out a network and you could click on a device and launch an ssh client.

We do lots of labbing of new networks for customers and it would be really handy to have a similar bit of software - does anyone know if anything like it exists?

I’m after something super simple that can draw a map and ssh on a click. It would also ping all the devices every 60 seconds and set of a horrific alarm noise if something went down.

Here’s a link to WS-Watch: http://www.cyto.purdue.edu/cdroms/cyto4/8_websit/valet/ws_wat1.htm

Best, Leigh



Route reflect newb - iBGP question

Hello all,

I'm looking to deploy iBGP with multiple route reflectors for redundancy. One thing I'm not quite understanding is exactly how the multiple RR works. Example: If i use 3 route reflectors will three copies of the routes be shared? I'm trying to determine what behavior to expect and how to set "primary RR's vs backup"

​

Sorry for the newb question just a little hung up on this.. Any input would be appreciated.



Multiple Subnets Single Cisco ASA Internet Packet Tracer

I was wondering if someone can help me with my packet tracer config.

I am trying to configure the following

Site 1 10.30.x.x >Directly Connected to Site 3 (Using EIGRP)

Site 2 10.31.x.x >Directly Connected to Site 3 (Using EIGRP)

Site 3 10.32x.x >Directly Connected to Site 1 + 2 (Using EIGRP)

Default Route for Site 3 Pointing to Cisco ASA

I can ping from Site 1 to Inside interface of ASA and vice versa

I can ping from Site 3 L3 switch to Internet (8.8.8.8)

I cannot ping from Site 1 to outside interface of ASA or 8.8.8.8 (web server)

Any help is greatly appreciated

Packet Tracer file at this link https://file.io/Uz1rj0UPkmYi



Online IPv4 NAT Service?

I've to run 13 instances of a service which can only handle IPv4. Each service must be reachable via the exact same port from the internet. This is a limitation which I cannot get around because I do not have control over the other side of this service. Thus I need 13 public IPs. Unfortunately my provider is not willing to provide me with a /28 IPv4 subnet.

Now I was wondering if I could get around that by running a firewall in the cloud, setup a VPN and book the needed IPs in the cloud and expose the services via the VPN to the internet. Has anyone done something similar or got a better idea?



Fastly CDN Outage, many sites offline (Reddit, Twitch, GitHub)

Since the Americans are asleep and nothing has been posted, I'll start a thread on the big news of today: large parts of internet become unreachable due to Fastly CDN outage.

https://status.fastly.com/

https://techcrunch.com/2021/06/08/numerous-popular-websites-are-facing-an-outage/?guccounter=1

https://www.theguardian.com/technology/2021/jun/08/massive-internet-outage-hits-websites-including-amazon-govuk-and-guardian-fastly

At the moment there's not much info, apart from a "service configuration event" caused an outage of the CDN Edge services that Fastly provide for these various websites.

Everything is back online as of 10:57 UTC, but websites may experience load issues as they back ramp up to normal levels.



Fiber redundancy network design

Hi guys,

i'm a little stuck with a fiber redundancy network design, it's kinda hard to explain, but i'll try.

Imagine:
-2 sites, with 2 firewalls on each side (VRRP clusters)
-Multiple direct fiber connections between both sites
-Currently, only 1x fiber is connected via a single fiber2copper converter
-On site A, the currently active fiber is directly connected to the local core switch (fiber converter A)
-On site B, the currently active fiber is connected to the firewall VRRP master (fiber converter B)

What i'd like to do:
1) Connect all fibers (current and new), from firewall cluster A to firewall cluster B
2) Create fully crossed redundant paths with all existing fibers between both sites

Now, you could create a new VRRP pair on both firewall clusters and simply connect site-A and site-B with all available fiber connections, but VRRP would probably not work, since master and backup can't see each other over the fiber links.

firewall-master-A
firewall-backup-A
| | | |
| | | |
----------------------------
| | | |
| | | |
firewall-master-B
firewall-backup-B

A workaround would be to drop the VRRP idea and work with watchdogs and routes, but that doesn't feel right.

Another idea i came up with, is the following:
-Each site gets a fiber switch stack, with SFP LACP LAGs (master and backup) for all fiber connections,hence creating a redundant layer 2 broadcast domain between both sites.
-Site-A firewall cluster will be cross-connected to the fiber switch stack A
-Site-B firewall cluster will be cross-connected to the fiber switch stack B
-Create VLAN(s) and STP on both fiber switch stacks
-Finally setup new VRRP, subnet, routings, inbound/outbound filtering on both firewall stacks

firewall-master-A
firewall-backup-A
| | | |
fibre-switch-master-A
fiber-switch-backup-A
| | | |
| | | |
---------------------------------
| | | |
| | | |
fibre-switch-master-B
fiber-switch-backup-B
| | | |
firewall-master-B
firewall-backup-B

I don't know how you would call something like that, i would call it a "transit network" (?), yeah, i just made that up. Not sure if the fiber switch idea is cool, stupid, or reasonable but overly complicated.

I'm not looking for a solution in particular, but rather a best-practice or a hint on how multiple redundant fiber connections between remote sites is usually done by the network pros?

Any inside is much appreciated.



Strange Cat5 cable speed issues

We have a client who was not getting the full 400Mbps speed from their internet connection. Upon investigation, we found their Unifi Secure Gateway was auto-negotiating the WAN connection to the ISP router @ 100Mbps.

We put this down to a poor cable so attended the site to replace the cable.

I initially tested the ISP's router with just a short pre-made cable and whilst I wasnt getting the advertised 400Mbps we were getting 200+ Mbps.

So I made a 30-metre cable round to where our USG was and before connecting to our USG I tested with just my laptop to ensure I was still getting the 200+ Mbps. Nope, it was hard capped @ 94Mbps.

Went back to the ISP router with the pre-made and tested and sure enough, 200Mbps again.

Thinking it could be the quality of cable I was using I made a shorter 2m length of cable and tested just into the ISP router and again 200Mbps

Ok now that's weird

So I then thought the length of the cable, or something I have run it next to is causing issues, so I pulled the cable back, laid it across the floor and chopped about 10m off it, tested and again hard-capped @ 90Mbps.

Any thoughts as to what could explain this?

Just to rule them out:

  • I tested on 3 different speed testing sites each time and all showed the same results
  • It was the same port on the ISP router each time
  • The ISP router is a high-quality Cisco router