Graham Beneke

@bettersafetynet @slippyfox Live migration is definitely possible on Proxmox. It does have dependencies on the storage being used. Ceph is the advanced option if you're doing complex clusters with a large fleet. A simpler option is to just use ZFS for your local vmdisk storage.

@slippyfox For lab/testing, proxmox is amazing. I've had mixed experience with high availability. Mainly it's rather hard. :-/ AFAIK, prox doesn't have an equiv function as VMotion. No, ceph doesn't count. (or prove me wrong prox fans!)

Dear, All those folks who were telling me I was wrong when I said VMWare/Broadcom would take away the free tier. History (especially with Broadcom) is a VERY good predictor. I'm sorry you're going through this. I'm not even saying "I told ya so"... 1

@TracketPacer In a joking manner, fiber is life! The literal answer youtu.be/mWqe8_5SUvk. It's surprising how much one can learn from this talk.

@mrhameljr @TracketPacer Richard updates this presentation every few years and it's an excellent resource. I've made every new team member watch it... Multiple times over if necessary

i tried to find the right SFP+ for something the other day for the first time in my entire life & honestly it was a bloodbath. the fuck is fiber anyway

@nickhedley @JK_ZA The lower stages of load shedding due to the Steenbras Hydro Power Station probably reduce the demand for solar installs in CoCT. Cape Town users experience minimal inconvenience from load shedding up to (Eskom) stage 3

@JK_ZA Yip. Although the Cape should be further ahead given its feed-in tariff incentive. It's clearly not ultra successful

Gauteng is comfortably South Africa's rooftop solar leader, with 1.3GW installed on homes and businesses around the province. For context, the Western Cape has only half that amount, according to Eskom's estimates.

@nickhedley @Arfness Thyspunt in the Eastern Cape was announced a few years back. There still seems to be ongoing activity by Eskom to obtain a nuclear operating licence from the regulator for the site. Multiple fumbles though.

@Arfness KZN would be my bet. The ANC in KZN has already said it wants nuclear

Anyone wanting to develop grid-scale power projects in South Africa is now confined to the yellow and blue substations. The grid in the rest of the country has no more capacity.

We do 32GB standard. 🤷‍♀️

@troyhunt @awlnx @haveibeenpwned Companies I've worked for have a process for that, but also load it up with so much bureaucracy that staff choose to just absorb the cost. For a few meals it's no problem, for a regular payment its a perverse incentive to push bureaucracy on others to avoid more of your own admin

@awlnx @haveibeenpwned What happens if they travel and need to buy lunch on company time?

We often receive comments to the effect of “we want to purchase a @haveibeenpwned subscription but our company doesn’t allow us to use a credit card”. What is the financial reason behind this? This is a very small portion compared to those that *do* pay by card, but why is this?

@adrianfrith Looking at the population age graphs from the previous census there seems to be a weird spike at 25-29. Is this something that is easily explained or something we need to watch?

For those who might be wondering: Stats SA is releasing the Census national results on Tuesday, but that's only the high-level figures. I will only be able to update dotmap.adrianfrith.com and census2011.adrianfrith.com when the detailed data comes out in a later phase.

@dtemkin Did some napkin maths around this for an FTTH network to look at the risk Starlink posed to business: Each Starlink satellite can support 2000-5000 households with a fibre-like service. Even at suburban density they could not launch enough satellites to displace fixed line

@erikbryn @GalileoSats Why does this profile correlate with the Northern hemisphere summer? Is the average being calculated using more Northern data points? Is the planet warming more overall during Northern summer? Something else?

You may have seen this, but I bet all your followers haven’t yet. They should.

@meileaben @atoonk @heymingwei Many of the IXs are running their own peering LAN systems on virtualised infrastructure. If the collector can be packaged up as a VM image then it could certainly be abstracted away from the physical hardware and logistics challenges.

@atoonk @heymingwei RIS has a mix of IX and multihop collectors (RRC00,RRC25,RRC25). Not everybody will do multihop, for multihop you will have to collect peer location metadata and at IXes RIS get feeds we don't get otherwise. But I get your point on the challenges that come with running at an IX.

RIPE RIS rrc19, hosted at NAPAfrica IX, will be shut down due to hardware failure that could affect data integrity. New updates files have already stopped showing up on the archive. No timeline on recover at this point. #BGP #MRT ripe.net/ripe/mail/arch…

@atoonk @heymingwei Using multi-hop has other challenges: you're dependant on a lot of intermediate networks and you'll lose connection and data streams during disruptions - which has historically been some of the most valuable RIS data.

@heymingwei I never understood why these collectors have to run physically at the IX, with all the challenges that come with that (power, space, hardware, shipping, scale limits). Seems the multi hop option to some (set of) vm’s somewhere is much better / easier

@TracketPacer In the past many switches would drop them - but then came Q-in-Q and dotQ tunnels. For a while this had to be explicitly enabled on the ports. These days the default config on many devices will just add the tag specified on the port and carry the resulting double tagged traffic.

i’m curious about this as well!

@woodyatpch @GuyHalse can you help with GINX? The online record is almost completely gone

Hey, PCH has been supporting a couple of historian-of-technology / archivist folks to build an indexed public archive of primary sources relating to the establishment of early IXPs. It’s all going into the Internet Archive, and a couple of other public archives, so anyone will be able to refer to it in the future. We’ve had quite a bit of help from folks who were participating at the time, but more sources and participation always makes for a better result, so I’m soliciting all of you who may have old emails or meeting transcripts or other documents, or know where there are public records or press clippings that pertain to early IXPs. The project is mostly starting with the 1992-1996 timeframe, but there are “early IXPs” in every region… for instance, IXPs started reaching the Caribbean mostly in 2011-2015. Eventually we’d like to see documentation of the establishment of every IXP, if interest in the project continues. So, if any of you are able to help, please contact me, and I’ll put you in touch. I really appreciate any help or information anyone can put toward this project.

We are pleased to announce that our next #community run #IXP will be the Nelson Mandela Bay IXP (NMBINX) in Gqeberha, Eastern Cape. Go live will be first week of June. If you're new to #peering be sure to attend one of our training sessions, to make sure you're ready to peer!

@MalwareJake Ignoring the license USB... What is the equivalent for read-only media like DVD in 2023? Something which is not going to raise red flags in air gapped environments.

The year is 2023. Siemens is still distributing software via DVD. And yet we know they know what USBs are - they include your license key on one...

youtu.be/KmfhhvFUYGY @GrahamBeneke For those who missed Mr Graham Beneke's presentation on "Scrapping 3064s... and other Layer-2 best practices" at #ZANOG23, we've got you covered!

@danieldibswe If your network availability target is in the three 9's (99.9%) ballpark then stacking is probably fine. It severely limits your options for concurrent operation during many types of maintenance. You should not be leaving a device for years with no software updates anymore.

In the networks you work on, do they allow stacking in the Core layer? Pros - Single management, less reliance on STP, bundling of links, no need for FHRP Cons - Potential SPoF, upgrades more difficult What do you think? #CCDE

The burden and privilege of leadership is accountability. Sadly most people think leadership is having the power to blame subordinates for your mistakes.

Mandatory password changes are stupid! I’m now locked out of my computer because of one. Please explain how this helps security?