infrapedia 2.0

Hello everyone,

I hope everyone is doing in these crazy times.

I am delighted to announce Infrapedia 2.0 which is a complete code overhaul from the previous version of Infrapedia. You can visit infrapedia 2.0 and published open-source code

For those who missed the original announcement, what is Infrapedia?

A tool that visualizes network and data center information and helps create custom views integrate with websites, monitoring tools (soon librenms), etc. One of the long time projects we have been working on is near-real-time crowdsourced network availability data. (think about Waze for fiber cut, ix outages, and other important information) (not focusing for home users but rather a long haul, metro fiber networks, ixps, dcs)

We’ve had to rewrite our code in order to be able to give new abilities to users such as integrating with your monitoring systems to overlay information, connecting owners of the fibers with buyers without middle-man, give the ability for people to use the data in custom formats for research as we did with Northeastern University,

we’ve created a slack to continue discussions and get more direct feedback from everyone who is interested, you can join our slack by clicking here

I appreciate all of the volunteer’s time who spent the time providing feedback, make recommendations, and more.


Good afternoon.

I'm looking for a switch with the following capabilities:
. transport for more than 20 gbps
. link aggregation LACP
. slots for SFP+
. seamlessly when trunking vlans through the link aggregation.

And essentially that doesn't exceed US$D 2000 and is compatible with 10GBASE-ER and/or 10GBASE-ZR modules that are not from the vendor itself (e.g. SPFs: Huawei, Mikrotik, Sumitomo, OEMs).

If any of you have a good experience with a device that meets these requirements (which are minimal with the exception of price and compatibility) ?

Mauro Gasparini

Used Juniper QFX5100-48T will do it. Probably overkill but you can grab one cheap @ebay

Hello Mauro,

the ICX7250-24P-2X10G may fit. 2×10 GbE SFP+, 6x1 GbE, can be licensed for full 8x10G.


Thank you. The problem is that to get a price lower than U$D 3000 I have to resort to a used device.

Have you actually looked at Mikrotik switches? I don’t like the OS, but the hardware does what you want it to.{%22sfp_plus_interface%22:{%22s%22:%223%22,%22e%22:%2224%22}}#!

If necessary, buy your SFP modules from and get them coded as Mikrotik modules at the factory - that’s what we do for Cisco, Brocade, Juniper, Extreme, etc.

Even the top-of-the-line Mikrotik only costs US$899.


While I am not sure on the optics support - what about a Ubiquiti EdgeSwitch 16XG?

12x SFP+ ports
4x RJ-45 Ports

It isn’t expandable - but maybe 12xSFP+ is enough?


Nathan Babcock
President - Straight Shot Wireless

O: 575-208-7915

Straight Shot Wireless


Mikrotik is a decent product and I’m always amazed at their features set, but…

Using the UI to configuring the switch will punt too much traffic toward the CPU. However, it is possible to configure the switch fabric with the cli and attain the desired results.

It does not compare to a JNP QFX5100 or a Extreme Network x650, x670… etc.

Thanks. I am currently using CRS326-24G-2S+ (with routerOS) for this, but when I want to do vlan trunking through the bonding, cpu load grows from 3% to more than 90%.
I’m consulting Mikrotik’s support in the meantime, because routerOS probably doesn’t take advantage of the switch hardware as swOS would.

I have no experience with Ubiquiti EdgeSwitch. I know its wireless product very wel, but I have no knowledge of how switches behave.

It doesn’t matter whether you use Winbox, web interface, or CLI, the same configuration capabilities exist.

The determination is how you configure it, what version of RouterOS you have, and what the hardware you have is capable of. The CRS3xx series are the easiest to configure, with the older CRS and switch chips in various other products of theirs getting progressively more difficult and more limited in hardware-accelerated features.

Yes, cpu loads a little more only when you are logged. The rest of the time is the same.

This works well on my CRSs:

/interface bonding
add mode=802.3ad name=bond-inet slaves=ether9,ether10,ether8 transmit-hash-policy=layer-2-and-3

/interface bridge port
add bridge=br-cabase interface=bond-inet
add bridge=br-cabase interface=sfp1

But if I want to bridge vlans behind some bonding Instead of bridging phy interfaces, cpu explodes:

/interface vlan
add name=vl201-mmen vlan-id=201 interface=sfp1
add name=vl201-mment vlan-id=201 interface=bond-inet

/interface bridge port
add bridge=br-mment interface=vl201-mmen
add bridge=br-mment interface=vl201-mment

On the CRS 3xx line, use vlan filtering instead. This guarantees hardware offloading.

PS. Do not use this method on the 1xx or 2xx lines.

/interface bonding
add mode=802.3ad name=bond-inet slaves=ether9,ether10,ether8 transmit-hash-policy=layer-2-and-3

/interface bridge

add name=bridge vlan-filtering=yes

/interface bridge port

add bridge=bridge interface=bond-inet

add bridge=bridge interface=sfp1

/interface bridge vlan

add bridge=bridge tagged=bond-inet,sfp1 vlan-ids=201

It’s clear then that I must use “bridge vlan” to achieve the goal I am looking for.
Now it’s time for me to study, research and test on my side.
If I have any specific questions, I will draw on your experience.
Thanks a lot.

That’s a downfall of Mikrotik, they give you ultimate power. You can do some pretty atypical things on there.

Yep, run SwichOS, prevents you from running things in software. :blush:

I could never get LACP + tagged VLANs to work on SwOS.

Then again, it doesn’t work reliably on RouterOS either, so I gave up. Spending more on hardware that is well supported is worth it versus my time and sanity.

I think Ubiquiti pretty much has the “cheap hardware that works well, but commercial support lacking” market cornered.