Post Azgo5TwLLOjRaSN8Gu by liaizon@social.wake.st
(DIR) More posts by liaizon@social.wake.st
(DIR) Post #Azdib7IdsDYRR8ck2S by liaizon@social.wake.st
2025-10-27T15:59:09Z
1 likes, 1 repeats
Is there a good article or thread countering this narrative that Signal (and communication of that scale) is not possible without its relience on AWS? https://mastodon.world/@Mer__edith/115445701583902092
(DIR) Post #Azdib8W9LTxBDJx2jQ by liaizon@social.wake.st
2025-10-27T16:03:59Z
0 likes, 0 repeats
I get the point they are trying to make, but don't think I agree with the premise. And the fact that Meredith mentions "Mastodon" in the list of names that rely on them: "Which is why nearly everyone that manages a real-time service–from Signal, to X, to Palantir, to Mastodon–rely at least in part on services provisioned by these companies." https://mastodon.world/@Mer__edith/115445706338981041
(DIR) Post #Azdib9cZF6gIdVxgNE by aral@mastodon.ar.al
2025-10-27T16:13:34Z
0 likes, 0 repeats
@liaizon She’s likely mentioning Mastodon due to S3 support for media. And once you’ve set up media in one place it is not possible to migrate without breaking existing links as they are hardcoded instead of being aliases. So it is a fair mention (as are your points in the follow-up post).
(DIR) Post #AzdibHKQaQvOXZXJD6 by liaizon@social.wake.st
2025-10-27T16:09:16Z
1 likes, 0 repeats
"Mastodon" does not rely on hyperscaler providers cause every single instance has the ability to move hosts if they choose, even if most servers are using OVH or Hetzner they can move from their own choices. And also Mastodon doesn't rely on Mastodon cause a large part of the "Mastodon network" isn't using Mastodon at all.
(DIR) Post #AzdinAO8DSneUTwAnw by cadey@pony.social
2025-10-27T16:15:47Z
0 likes, 0 repeats
@aral @liaizon There are ways to do the migration more gracefully fwiw, it's just a pain.
(DIR) Post #AzdjbZmI5lwfiMH5ou by aral@mastodon.ar.al
2025-10-27T16:25:04Z
0 likes, 0 repeats
@cadey @liaizon Good to know (not least of all because it’s one of the last things tying me to Bezos). Time… now I just need time! :)
(DIR) Post #AzdypeLBQIXvESlRU8 by shalien@mastodon.projetretro.io
2025-10-27T19:16:14Z
0 likes, 0 repeats
@liaizon And what you're calling mastodon is just a software built upon the Activity Pub protocol which is the key element in doing all of this
(DIR) Post #AzeXR5pXZqCde6zPgu by teajaygrey@snac.bsd.cafe
2025-10-28T01:43:25Z
0 likes, 0 repeats
Many seasoned ops folks such as I have been warning people to avoid "the cloud" for what feels as if it is now decades.Largely, it seems as if we are ignored?I dunno, you want a nicely packaged article or thread countering sage perspective from experts, you might not find it.Maybe something like this about saving millions by leaving AWS: https://www.theregister.com/2025/05/09/37signals_cloud_repatriation_storage_savings/ ?Problematically, even the term "DevOps" is offensive AWS peddling trash to the likes of me.Which is to say, to someone such as I? The "Dev" in Ops is silent. If you give folks root/admin/enable/etc. access and they are not already superlative developers? Someone fucked up somewhere.To put things in perspective, writing personally: this current human incarnation of mine began in Menlo Park, CA in the 1970s. I was exposed to systems some only ever read about in history books as a child (think: GUIs with mice and such before the Apple Lisa was even a thing).I took my first formal course in programming when I was six. By the time I was a teenager, I was getting invited to toil on systems at nps.navy.mil (now nps.edu; aka the Naval Postgraduate School. I was unpaid, and uncited; what today might be termed by some as "shadow IT" but mostly just grateful to get to use some Sun Microsystems workstations and even better: Silicon Graphics workstations, one of which cost more than the mortgage on my parents' home at the time) and was even told that some of my code ran on the $30 million Cray at FNMOC (Fleet Numerical Meteorology and Oceanography Center, this was in the early 1990s, before Cray and SGI merged) though they never let me "get time" on that or even touch it, I was only allowed to look at it from behind plexiglass.By the time I graduated University, I got a job at some "pre-IPO" SS7 to web telemetry provider that had roaming contracts with every carrier in North America as a (Lead) Network Technician and I helped drag their sorry asses from "two nines" of uptime per annum two "three nines" (also see: https://en.wikipedia.org/wiki/High_availability#%22Nines%22 where "five nines" was considered industry leading).By 2009? My then employer (where I was a "Senior System Administrator") was still up and passing packets happily, even when the fiber ring in the SF Bay Area was cut in various places, so bad that AT&T (one of those supposed "five nines uptime" telco providers) customers could not even make cellular calls. (article on that outage: https://www.wired.com/2009/04/cable-sabotage/)All of that was done: without AWS.My employer circa 2010-2012 had Facebook, Micro$oft, Google, Bank of America and more as clients. I was distressed to learn how egregiously wasteful most of those companies are with regards to computational resources.In contrast, the employer I had off and on circa 2007-2014? Which had been online longer than Google? With one of the oldest affiliate systems (albeit "adult" in nature) had maybe 1/10000th the number of systems in operation as say, Google or Facebook (last time I heard metrics on their deployments they had something like 3 quarters of a million servers).Which is to say: if you know what you are doing? You don't need a lot of resources.If you are fucking morons? Then yeah, sure, you can waste inordinant amounts of computational resources and money on so-called "hyperscaler" bullshit being peddled to you by "DevOps" hook line and sinker shills for AWS and the like.Heck, I was a Senior System Administrator in 2016 for a company that ran the cross-browser development infrastructure used by Fortune 1, among others. We had approximately 40 racks of gear spread across 2 data centers (which, if you say have 42U racks and each system were 1U [most weren't that dense, some blade systems notwithstanding would imply we had at most approximately 1680 systems. We had far fewer than that in reality, though the number of VMs we spun up was significantly more than the physical hardware we had), with plans to start a point of presence in the EU. That employer did have some AWS usage, and our small ops team was hell bent on eviscerating it because we had faster, cheaper, local storage already and the AWS dependent dev was running up something like $10k/month on what was really not much data and we could be buying NAS devices, plural, which would accommodate their needs for that kind of burn rate.Ironically, that first employer I had after I graduated from University? Seemed to have gotten dumber somewhere along the line, I did an evaluation of all my past employers' networks for shits n giggles, and noticed they were relying on AWS, and were down during some AWS outage. Which means, that those "three nines" I and my 50 some odd coworkers were so proud of back in 2000 or whenever that was? Had dropped back to the "two nines" or lower levels of "reliability" since that company migrated their shit to AWS instead of keeping it in house.Dumb fucks, are fucking everywhere.
(DIR) Post #AzeXuvqZGAGJN1hM8W by teajaygrey@snac.bsd.cafe
2025-10-28T01:48:50Z
0 likes, 0 repeats
If you want more "concrete" examples though than my own navel gazing?Consider, DNS.The root name servers, are not centralized, by design.I would trust DNS a lot more than anything from AWS or Signal.Heck, that's even after I helped patch https://nvd.nist.gov/vuln/detail/cve-2013-4854 while I was consulting at iXSystems and it was still under embargo (not publicly, but being actively exploited).Most well designed Internet "scale" technologies are similarly well thought out like DNS; SMTP (aka email) for example, it's common to have backup MX servers (plural). Most firewalls I've been deploying since the 1990s have various forms of fault tolerance (and thanks to folks like Ryan McBride and the OpenBSD developers we now have CARP [Common Address Redundancy Protocol] instead of needing to rely on vendor proprietary closed source things such as VRRP and NSRP). BGP, similarly, facilitates stability with routing, etc. etc. etc.There are a lot of reasons not to trust Signal, over reliance on AWS isn't even one of the big ones, but is just another reason to add to the pile of reasons to avoid Signal.
(DIR) Post #Azef9BFdU0EwsaAI4G by tomjennings@tldr.nettime.org
2025-10-28T03:10:25Z
0 likes, 0 repeats
@liaizon Mastodon doesn't need to manage isochronicity at the milliseconds level. I *trust* the Signal folk to do their own right thing. Which isn't the same as "believing".
(DIR) Post #AzegTaOpn8HCcalHDE by liaizon@social.wake.st
2025-10-28T03:25:14Z
0 likes, 0 repeats
@tomjennings the argument comparing Mastodon to Signal is weird and not my own point here at all, it came about cause Meredith had mentioned Mastodon in her thread. As someone who has worked on stuff that was used by a huge population, do you think its not feasible for them to remove their reliance on GAFAM?
(DIR) Post #AzejJhGtgco2tw1cHI by liaizon@social.wake.st
2025-10-27T16:31:53Z
0 likes, 0 repeats
I would be interested to see some statistics for what different VOIP providers are using now, they can't all be using AWS, I wonder what some state services are using internationally
(DIR) Post #AzejJiJPokPm82D8qG by liaizon@social.wake.st
2025-10-27T16:55:23Z
1 likes, 0 repeats
Glad that Signal exists and has paved the way for encryption to become as mainstream as it is. But they also need to get off their high horse. There is so much to be done in this area and it's sad to see how little focus some of the "smaller players" get in this respect.
(DIR) Post #AzejJqL7yGXP1f4cca by liaizon@social.wake.st
2025-10-27T16:59:32Z
1 likes, 0 repeats
I am not a professional in this space. I am simply a human trying to understand the world I live in. Signal and the people who represent it have continually dragged their feet in so many areas. The fact that you still need a fuckibg phone number to use Signal is a disgrace. Yes finally you can mask it with a username, but you still need access to a phone number to register an account and the amount of unneeded hastle that causes is immense
(DIR) Post #AzejJzVNtV5dS4wRMG by liaizon@social.wake.st
2025-10-27T17:02:14Z
0 likes, 0 repeats
Telegram is still winning in many regards. Why is there not 1 to many broadcast groups and more advanced group moderation features in Signal. Telegram is shit in many ways but it's usability wins most of the time.
(DIR) Post #AzejK8m1R0kKCHnM2a by liaizon@social.wake.st
2025-10-27T17:14:36Z
0 likes, 0 repeats
Also: "The CEO of the Signal Foundation, Brian Acton, has a reported salary of $785,000 per year. Additionally, the highest-paid software engineer at Signal earns around $489,000 annually, indicating that executive compensation at the organization is relatively high for a nonprofit."
(DIR) Post #AzejKGshI4q5LIynYW by liaizon@social.wake.st
2025-10-27T17:16:42Z
0 likes, 0 repeats
This is just DDGs auto generated answer so I can't be sure this is true, but what I do know is that Signal has a CEO. Our global communications SHOULD NOT HAVE A FUCKING CEO. AND THEY SHOULD NOT BE BASED IN FASCIST USA
(DIR) Post #AzejKP8aasIwwuTbRA by liaizon@social.wake.st
2025-10-27T17:19:51Z
1 likes, 0 repeats
Ok so this is some BULLSHIT right here: https://projects.propublica.org/nonprofits/organizations/824506840Thanks to @ProPublica for publishing this data
(DIR) Post #AzejKYCStpkj3XMKEC by liaizon@social.wake.st
2025-10-28T01:08:25Z
0 likes, 0 repeats
in 2023 alone the Signal Foundation spent $4,250,453 on the salaries of just 7 people. And this is suppose to be a non profit. They ask *us* for donations. (these are just the posted numbers from 2023, no idea if they have gone up or down since then)
(DIR) Post #AzejKhGLCnCVAAF31E by liaizon@social.wake.st
2025-10-28T01:18:38Z
0 likes, 0 repeats
they state that part of their mission is to "investigate the future of private communication" and it also seems part of that mission is to continually state that centralization is the only real option
(DIR) Post #AzejKpqRISpTmRBzOa by liaizon@social.wake.st
2025-10-28T01:30:59Z
0 likes, 0 repeats
reading the bios of the people listed on the Signal Foundations website... Katherine Maher:"She is an appointed member of the U.S. Department of State's Foreign Affairs Policy Board, where she advises the Secretary of State on technology policy." and "a term member of the Council on Foreign Relations, a World Economic Forum Young Global Leader"
(DIR) Post #Azeo2FdjfOMC3ECy3s by tomjennings@tldr.nettime.org
2025-10-28T04:50:01Z
0 likes, 0 repeats
@liaizon Yeah I read that too. It seemed an odd mention. I think I sounded more defensive of status quo then intended. It's just that at this point, here in the US, I'm tending to think differently about comms and their survivability.Signal is what it is. I do trust them, but they have this very large semi central system that has features, and vulnerabilities, and those won't structurally change in the short term. And no one knows if we're gonna have a long term here, just now.
(DIR) Post #Azes7NKYtq7VATwG36 by liaizon@social.wake.st
2025-10-28T05:35:44Z
0 likes, 0 repeats
@tomjennings if you were tasked to design a realtime secure coms system for the whole world starting right now where would you start?
(DIR) Post #Azf8jUSISNxdXHkkZE by cnx@awkward.place
2025-10-28T08:41:56.202045Z
0 likes, 0 repeats
@tomjennings, you don't need sub-500 ms latency for instant messaging, even if mega cloud providers are the only low-latency option for decent VoIP UX, there's no reason to not have reliable instant messaging when they're down. Whittaker was basically saying because frozen meat is cheaper, even if the cost of rice is negligible, when the supermarket closes we should rather starve than buying rice at the farmer's market. Say you're an activist, if you can't call for rescue you should get killed instead of sending out an SOS text. TBH it sounds like a bullshit excuse to continue cutting corners.BTW @liaizon, round trip time between data centers are up to 70 ms because physics, though realistically communication are usually between closer people. The majority of latency is between households and data centers due to the lack of network neutrality, but even that is usually only an issue for international connections, and 50 to 500 ms is a steep jump. In 2023, Signal had 55M users and spent $11M on infra (not counting cellular registration). That's one good rapper per user. Splitting that into 1000 servers (~5 per country/region) you will get a monthly cost of $1000 serving 55k users. If each is calling 10 min each 16 h day (because most people sleep at the same time) on average, that's 600 simultaneous users, accounting for around 10 Gb of bandwidth. That's still huge, but since Signal does not do homomorphic video processing, servers are just relaying the data.The math here are highly approximated with error in an order of magnitude, but it is clearly possible to use multiple infra providers. Cc: @rysiek and @lrvick to hopefully correct some of the stuff I got wrong here.
(DIR) Post #AzfKiaFsbhCkgSHNbs by maphouse@mapstodon.space
2025-10-27T17:21:42Z
0 likes, 0 repeats
@liaizon what are your thoughts on @delta ? I've found it to be the most exciting chat technology lately
(DIR) Post #AzfKib5zTwt9I4ezpo by liaizon@social.wake.st
2025-10-27T17:25:46Z
0 likes, 0 repeats
@maphouse @delta they are awesome and I am happy they exist, and I consider them my friends. Still a long way to go before they are competing in the same bracket, but they are getting there and making amazing progress with 1/100th of the resources. They recently started beta testing voice and video which is very relevant to this discussion.
(DIR) Post #AzfKibtyU6s3n62ukC by ati1@mastodon.social
2025-10-28T10:46:37Z
1 likes, 0 repeats
@liaizon @maphouse @delta Delta voice / video work just fine for me. Already live - tested multiple scenarios like Android to Linux phone. Amdroid to Linux desktop etc. Another one of "it just works as intended no hussles" for DeltaChat. I already have one relative who constantly calls me via DeltaChat video call because "voice quality is noticeably better than when I just use gsm audio connection" 💁♂️️
(DIR) Post #AzfLgvHJZYBnGEiqjA by raucao@kosmos.social
2025-10-27T16:03:40Z
1 likes, 0 repeats
@liaizon In short: her entire point is based on the fact that they want to be one centralized provider servicing people everywhere on the globe. Federation with many local servers would obviously scale just as well for the vast majority of their users.
(DIR) Post #Azfm0u2GV4Jq0Ss3UW by tomjennings@tldr.nettime.org
2025-10-28T16:02:05Z
0 likes, 0 repeats
@liaizon I would start by fundamentally rethinking what "whole world" means. The current idea is stupid and harmful.
(DIR) Post #Azgo5T6aRpKczw9nbE by beadsland@hcommons.social
2025-10-28T00:39:32Z
0 likes, 0 repeats
@liaizon They're talking about the entire technology infrastructure, not just backend storage:https://mastodon.world/@Mer__edith/115445707420096286"the entire stack, practically speaking, is owned by 3-4 players"How many users of fedi are not dependent on Apple, Google or Microsoft to read and post on a fedi client?AWS, per this argument, is a symptom of a much broader problem, that one doesn't escape by crowing about individual instances doing as they choose, because to access those instances, one still has to go through access providers.Now, one can argue that they're conflating AWS (backend) with device/OS (frontend) providers in a disingenuous manner. And there is some validity in that counter-argument.But that's why they're lumping Mastodon in there with everyone else. Because in the end, if you're using one of three device architectures as your access point, you're still dealing with hyperscaled (industrial monopoly) providers for that access.
(DIR) Post #Azgo5TwLLOjRaSN8Gu by liaizon@social.wake.st
2025-10-28T00:51:30Z
1 likes, 0 repeats
@beadsland I realize they are talking about technology infra. And in a decentralized system like the fediverse or matrix or delta or email or... you don't HAVE to depend on Apple, Google or Microsoft. In fact this place is probably one of the few that a sizable portion of its participants are not using any AGM back or front end. Yes the tech giants have an immense monopoly on all sorts of layers of the stack. But if a system is designed in such a way as to not require them we are much better off