Thank you to everyone who attended the inaugural LAN all NIGHT Gaming Festival! Thanks to you, more than 600 food items were collected for the North Texas Food Bank and more than 20 people donated blood to Carter BloodCare.
We want to release a report on our network performance and internet issues some attendees experienced during the event.
We’ve identified the following issues that likely led to the internet performance issues. Rest assured, that all of these issues will be resolved by the next event.
- The hotel’s WAN switch didn’t have an open port we could use, meaning that all of our traffic was being NATted by our PFsense router and NATted again by the hotel, which can cause issues.
- The entire hotel’s internet crashed Saturday night and required a reboot of their entire stack, possibly revealing an issue with the integrity of their network.
- It was identified that some attendees brought their own switch, connected it to our network and used it for their gaming.
In light of these identified issues, here is what are doing for future events:
- Make it clear to attendees that bringing outside network equipment is prohibited. Bag checks will be instituted to ensure this is enforced.
- We will explore network-based solutions to identify parasite switches on the network and/or deploy port based authentication to ensure that once a port is taken by a PC, only that PC can use it, preventing anyone from using an outside switch.
- We will work with the hotel to ensure we can obtain direct WAN connectivity for the next event. Whatever steps we can take to help facilitate that, we will do.
- We will remind the hotel of the network crash they experienced during the event and encourage them to verify the integrity of their network prior to our next event.
Although the internet had some issues, the server performed just as expected.
“The cache server performed beautifully,” said Guy “Halestorm” Hale, LAN all NIGHT Network Engineer. “It was so fast, I imagine download speeds of cached content were only limited by the write speed of the drive in your PC.”
2.2 TB of content was downloaded to the cache server and 6.2 TB was served up. The max upload bandwidth recorded was 6.55 Gbps, so the 10G NIC was definitely being utilized.
- i7-5930k clocked at 4.3 GHz
- 64 GB of RAM
- 10G NIC uplinked to the core switch
- All switches had dual 1 GB connections in a port channel to facilitate redundancy and extra bandwidth in the event a switch was saturated with downloads.
- The router and cache server ran as virtual machines under Hyper-V. I used PFsense as the router (4 threads and 2 GB RAM assigned, was overkill, we never exceeded 55% of the resources assigned) and we used the latest LANCache Docker container under an Ubuntu 18.04 LTS VM with 12 threads and 48 GB of RAM assigned.
- The cache server ran on 8x 500 GB Corsair MX500 SSD drives in a RAID 0 for a total usable cache space of 3.3 TB. Measured test write speeds were about 8 Gbps and read speeds easily exceeded 10 Gbps.
We appreciate all of you for attending and apologize for any issues you may have experienced. Rest assured we will work hard to ensure our next LAN is smoother than a llama’s backside!