Internet Disruptions & Iran’s Shutdown: Cloudflare Radar Insights (Storm in Portugal Included)

Hello everyone and welcome to this week in net is the January the 30th 2026 edition. This week we're talking about internet disruptions with a focus on the multi-week government imposed shutdown in Iran. The situation there is really difficult with serious humanitarian consequences and the internet shutdown only adds to this troubling reality. Also was worth mentioning in our blog this week. We had a very cool blog related to something that many are talking in in the industry uh in the tech sector. Uh it's um a self-host personal AI agent. So it's all about having a real AI personal assistant working for you. Uh initially this project was called uh Claudebot then it changed its name this week on Tuesday to Maltbot and now because the um the creator of of this project that is based in Austria didn't like the change that he did did uh for Moldbot now uh this Friday is already a different name it's open claw um so now it's open claw but this blog post uh explains explains uh how we're introducing what we call maltworker. Moltworker is a middleware worker and adapted scripts that allows running uh this open claw uh on cloth's sandbox SDK and our developer platform APIs. So you can self-host your AI personal assistant without any new hardware. So Apple minis have been really popular because of this AI personal assistant. This means you don't need a new hardware to run your AI personal assistant or even VPS as well. Uh we also had earlier this week a blog post about uh a proof of concept really uh about how we build a matrix home server to cler workers. So also a blog post about developers. So it's all about delivering encrypted messaging at the edge with automatic uh postquantum cryptography and also really as usual popular blog post because it was about an incident that Clover had uh and we're really transparent on transparent on these. Uh so it's route leak incident on January the 22nd. Uh an automated routing policy configuration error causes to leak some border gateway protocol prefixes unintentionally. So that's BGP pro prefixes from a router at our MA Miami data center. So we discussed the impacts and the changes we are implementing as a result. Of course being transparency really part of the culer uh culture. It also brings us a lot of attention. People really like to read these technical blogs usually and that's because we share actual data actual information with industry for others also to improve and be more resilient. Um this Friday it's coming out. So you when you're watching this, it's already came out uh a blog post about Google's AI advantage and why crawler separation is the only path to a fair internet. This is all about a new proposal from the UK about new rules to give publishers more control over how Google uses their content in narrative AI. But the measures from Cler's perspective still fall short. So a stronger separation between Google's search crawler and its AI systems is needed to ensure real transparency, fair comp competition and better protection for the creative sector. That's something that Crawford defends in in this situation. The UK proposal is already a good a good step there uh to that purpose, but we actually defend a bit more protection there and better separations. So without further ado, here's my conversation about internet disruptions with my colleague David Bellson. I'm your host Ronto based in Lisbon, Portugal. A very rainy and stormy Lisbon, Portugal. And with me, I have returning to the show, David Bellson, our head of data insights, and my colleague in the radar team based in Boston in the Boston area in the US. How are you, David? Where it is not raining, it is really cold and uh lots of snow. No snow. Oh, snow. Well, it's not snowing. It's not actively snowing at the moment, but this weekend again, we're supposed to get more. So, snow around. And we we had a clear a big storm this week here in not only in Lisbon, in Portugal. There was actually two regions in Portugal that were were more affected. That was Leia and Santi, but also Quimbra. So, actually three regions, but two were more affected. And this was the storm was in the early morning of Tuesday to Wednesday. And by 400 p.m. there was like a clear uh power but also internet uh disruption and we saw traffic going down as much as 70% in Laria and close to 60 uh% in the other two and dur mix I assume of of power outages and some infrastructure damage as well. Yeah, there was a lot of of both. Yeah. Yeah. And uh currently we're recording this on Thursday is still ongoing in Liia. So over 24 hours of disruption in terms of power outages, but also internet outages. Uh I believe there was 1 million households at at the beginning of the power outage that were affected. Uh so a clear impact here. Yeah, those those often take a long time to uh to resolve. Thankfully, I think we had so the snow I was referencing was from a storm last weekend that um kind of moved across the the midsection of the country and up into the east coast. And there was a lot of ice and a lot of snow across a lot of states, but you know, keeping an eye on the connectivity and the traffic levels from those states, we didn't really see any significant impacts. There were definitely power outages. There's definitely some infrastructure damage, but I guess it wasn't quite as widespread enough to to really register at least on, you know, as unusual and anomalous in our our traffic graphs. True. And we're not accustomed to these types of storms here in in Portugal. And the the winds were up to 95 miles hour. That's so many we're supposed to get that from this where they're talking about a noraster or a winter nor easter, which is often, you know, very cold, snowy, and high winds. So, it's the the triple whammy. Let's why not go directly to the Iran situation that started uh a while ago, several weeks ago. What can we exactly uh what can we say about that? It was a clear shutdown of a very big country in terms of population, in terms of area that was almost close to zero for many days, right? Yes. Yeah. So, Iran has shut down the internet in the past. So, this is not unexpected. uh you know they have a playbook for better or for worse on how they they implement these types of shutdowns and they're usually pretty orderly but it appears that in this case it wasn't quite as orderly uh and it was a little more of a scattershot approach. So you know ahead of the shutdown and the days ahead of January 8th we saw some changes uh around IPv sorry advertised IPv6 address space that a lot of that had disappeared. uh there were some shifts we observed around the share of traffic using HTTP3 and quick where that dropped. So that indicated that there might have been some some filtering going on, some some weirdness going on there. And then I think midafter afternoon around I think it was around 16:30 on January 8th, we saw traffic just started to drop pretty rapidly. And if you looked at the major autonomous systems, the ISPs there like Iran cell and MCCI and TCI and other retell and others around that time they all started seeing traffic just just drop precipitously. That was the initial part of the shutdown. Over the last three weeks, um traffic has careful with my wording here. Uh we've seen traffic return in sort of fits and starts. Last week, there was a little bit of a an increase we saw where it was at near zero. Uh and it had grown about X. So, it was, you know, the traffic we were seeing was 50 times higher than it was previously the past couple of days, but it was still super low. Um that only lasted for a few hours and then went away. But this week actually we've seen we saw traffic come back again late Sunday into Monday and then dropped again. The late Sunday into Monday recovery was at least for us uh similar peak levels that we saw pre-shutdown uh and then um that dropped uh uh late Monday and then picked up again uh on Tuesday. Uh and it's it's stayed uh well I say elevated uh over the last couple of days. So I think you know there is some connectivity that's come back there uh based on you know what I've heard from from people in the know there you know there's a lot of allow listing going on a lot of deny listing you can you know they're saying you know users can get to these domains but not these domains slowly slowly enabling I think from uh some of the fixed line providers you know but there's there's been a lot of you know as the the shutdown was occurring there was a lot of talk about Starlink and how Starlink was was helping keep people online but the government was aggressive ly going after Starlink users, trying to find them, trying to block the the signals, arresting and finding people for for actually using the dishes. So, it's been quite a situation there. Absolutely. Absolutely. One of the things that surprised me is first it was almost close to zero shut down for so many days and the impact in terms of uh all the country not only communications of course the reason why this was done was because of the protests geopolitical situation there where people are confronting the government there. So it was to avoid the protest and react to the protest which is common as you said but the impact for a shutdown government mandate shutdown for so many days in terms of economy in terms of relationship between universities and the outside world people right actually that reminds me that that one of the first recover you know quote unquote recoveries we saw was that there was a about five or six I think university autonomous systems that we saw traffic return from uh a few days I think after the shutdown began but that was very very shortlived as well. Exactly. And it's it's really surprising to see how is that possible in 2026 already and the impact that has in terms of the economy of the country even for the future. So many days it really has an impact and as you were saying we can see definitely that the return is a different return is it's not a full complete return. It goes down and up and right and even the shapes of the traffic graphs. So you know like like you were showing earlier the pre-shutdown graphs are very clearly there's a very clear pattern to them a very clear dal pattern you know you can very clearly identify like hey this is normal this isn't the graphs now over the last few days have not had such a clear pattern they've been much spikier yeah you you know you can see the weekend there where where it drops in between the the the first and the fifth labels you know it's a very very very normal pattern and then you know over the last few days here it's sort of come back and it's gone away and then really hasn't been a return to quote unquote normal usage even at a lower level. Okay. And there's many sites that are blocked still. Yes, only a few are enabled. Of course, people on the ground are trying to have things working where they can go around some of the filtering that Yeah, there's definitely some a lot of circumvention efforts going on. Some successful, some not so successful. It's it's it's quite um in in a in a historical situation interesting to see but also worries us as humans in 2026 to see. Yeah. No, absolutely. Absolutely. I mean part of this is also enabled by you know all of internet traffic or all of international internet traffic funneling through a single provider. Um so that does provide something of a choke point to enable them to implement things like this. True. And one of the things that we know also is regarding I think filter watch an organization watering Iran's internet traffic also reported that services like Google being chatbot are available only in a few areas I think right it depends so the Google transparency report site you can look at it you know at a given country and see a in normalized view of traffic for a given country some of the Google sites have come back so like think news sorry I know Google search is available there but I think Gmail Email access has not really returned yet. Google maps may be available. I I forget which were. Um they've definitely enabled or or allow listed um some of the Google services but but not all of them. In terms of the monitoring of a situation like this one in Iran. What are and you you wrote a blog post specifically on this. What are the the the takeaways the key takeaways we we can take from from this type of outage? As you mentioned, Iran is a country where this happens sometimes, but not for as long as this uh this has been going on, right? Yeah. This is decidedly unusual, certainly at a at a at a national level. I think we we've definitely seen long duration shutdowns in the past, but they've been much more localized um and and in some cases even harder for us to see. But at a national level, a 3we shutdown is is particularly unusual. You know I think that what other you know I think the concern I think is that other countries look at this and say ah this is possible ah this is effective you know but I think that it's it's really not effective because there are in many cases ways around it I think what needs to be um promoted is the wrong word but but sort of socialized is is as you mentioned the cost so you know for every day that connectivity remains unavailable in the country that has a real impact to business. It has a real impact to their GDP. Uh and you know, it has a real, you know, multi-tens of millions of dollars a day cost for as long as they're keeping the connectivity offline. It's interesting to see in our data uh that currently with traffic coming back in part but not a fully recovery. The there's almost a battle between people trying to access the free internet but also the government trying to block them into specific sites. There's actually a report from Iranian news organization that for example explains that Iranian cos are gathering in the dining hall of the tan chamber of commerce to access the internet and their activity there in terms of the internet is monitored by the government. So there are like purses of internet access. This reminds me in situations in Africa where people to access the internet they need to go to a specific town hall or area that's the area in the city where there's internet. Yeah. And in this case there's also been talk of I think that's what they call white SIM cards where people who have been deemed I don't know worthy for lack of a better term are given sort of special access to you know the mobile internet. You know, and this is to say nothing of the the work that they've been doing within the country also on the NIN, the national uh information network where they're attempting to create sort of their own national internet effectively, you know, replicating a lot of the the service, the chat services, social media type services and search services and things like that. My understanding is that the way the shutdown was implemented that got affected too. So it wasn't even like they were shutting, you know, we're we're shutting the gate to the, you know, to the to this the the w the walled city and everybody needs to just, you know, use what's inside now. People couldn't even get to that either apparently. And in in terms of this battle, you can also see that there's attempts to do like new blocking like create the local internet blocking a lot of things and we can see in our own metrics differences between the uses you mentioned IPv4 and IPv6 initially right there's differences because there's definitely different things in a protocol level. We can see differences in some of the the TCP tampering metrics that we have. So, you know, prior to the shutdown, we saw this mix of of what I call tampering signals. And then as connectivity started to come back over the last few days, that makes it shifted a little bit, which could indicate a shift in tactics where where certain things were being done before and now they're they're doing something slightly different. Makes sense. Where where do you see this going next? Uh still not a fully recovery uh internet to to Iran. Where do you see this going next? Unclear that it will ever fully recover. You know, I think that um you know, based on things I've seen in the news and folks that civil society publications and things like that, I don't think there's a lot of confidence that things will go back to normal, such as such as normal was pre-shutdown. You I think I think there will continue to be some amount of limitations in place, blocking, filtering, surveillance, you know, things like that. what I what I hope doesn't happen um you know I I I would really don't want to see other countries looking at this and and taking um you know taking lessons from it from the perspective of like oh yeah we can do this too you know the lesson they should take from this is we shouldn't do this it's a really bad idea absolutely and of course as you mentioned consequence are not only for the population but for businesses there's all sorts of education health finance yeah absolutely every everything is going to impact true um quite in in a way also it shows us the fragility of the internet and how it's used when it's a government needs to create an impact a specific impact on controlling their own population. Yeah. And I mean it's definitely the the latest evolution of when governments used to take over the press or or the broadcast media to control the messaging and now it's sort of hey we've got a social problem and we're going to implement a technical solution. uh you know unfortunately a technical solution is is when you have a hammer everything looks like a nail. We've seen sort maybe even shifting to the blog post. We've seen you know shutdowns used as a as a tool in the past in many cases. You know we've talked a number of times about the exam related shutdowns which are also a bad idea. Uh you know because they're trying to solve a very small problem with a very large effort or not effort but you know very a very large uh action and that has you know arguably maybe less of an impact because of the timing of it. you know, they're shutting it down between, you know, 6:00 and 8 in the morning or or couple hours, whatever. But still, you're taking down a a u you know, communications vehicle that impacts the whole country. Yeah, it definitely makes a difference. We were mentioning specifically the the difference and changes that are there are in the country specifically and why not show radar and the adoption and usage page for Iran here. There's definitely even on the browser usage. Yeah, you can see when connectivity sort of I don't I don't want to use the word stabilizes, but I'll use the word stabilizes. I think the the very noisy signal that you see there between the ETH and and whatever that is the end of last week or 20th. Yeah. Yeah. Um you know was was because there was such little connectivity that whatever we were seeing was sort of causing that noise. But it appears that connectivity has opened up or or you know connectivity to cloudflare at least has has kind of become a little bit more regular which enables us to see a pattern there that is more similar to what we saw pre-shutdown. True even in the protocols we can see definitely a difference still ongoing. It's not still similar to the pre uh shutdown moments even on the HTTP3 perspectives here and ATP1 specifically. It also shows the browsers and the the usage here that is different right also in the iOS front much different that was something we were discussing yesterday and that's really you know it's an interesting observation to see that over the last like week that iOS has spiked so much and you know the the thought there one one potential thought there is that the folks who have been given access you know lean more heavily towards an iOS using population you know make make of that what you will but uh you know when pre-shutdown when things were more normal you know users represented that looks like about you know 10% share now it's a higher share certain certainly spikiness so it may be the case that the folks who have been given access and allowed access are are more heavily skewed towards iOS users that's also a trend specifically let's move on to to Q4 2025 as usual you write always this uh this internet disruptions uh reports and Q4 was busy although Q126 is also starting really busy but Yeah. Yeah. Q1 started off with a bang this year. But what can you tell us in terms of the the main drivers here for disruptions in Q4 2025. So I think I guess sort of thankfully there was only one government directed shutdown that we saw which you know there are some quarters the sec usually like the the third quarter is usually particularly bad because we've got Iraq and Syria and others that are doing exams. So we have these you know multi-day multi-hour shutdown programs. In this particular case, it was Tanzania around protests around the election. This is something we actually saw in Uganda earlier this month as well where there was a multi-day shutdown. But, you know, so there's that there was that issue. A lot of it was the usual suspects, you know, digital Haiti getting hit with cable cuts that happens, you know, all too frequently. There was, you know, weather uh severe weather. I'm trying to think of what else we saw. You know, it it was, you know, submarine cable issues that hit African countries, African providers, power outages. So we were discussing this earlier uh in Portugal where where power outages can ultimately manifest themselves in internet traffic graphs. Equipment falls offline, users, subscribers fall offline. Uh that ultimately results in lower traffic volumes. In Kenya as well, typically there there are countries that are more these power outages are more frequent. We know Cuba, we know high, we know Yeah. Yeah. Haiti. Uh Nigeria. So Nigeria is interesting because they've had some some countrywide outages. So we didn't see one in in Q4. But I do see news alerts quite frequently about yeah you know another nationwide power outage in Nigeria and you know the country is completely you know complete blackout whatever and I go and look at our our traffic data and I'm like there's been no change. So I'm trying to figure out where that sort of disconnect is. Uh you know I think that the the news is maybe overstating the case but but yeah there's definitely a I should say this there's often a correlation between power edges and internet shopping. True. And of course military action in Ukraine. Yeah. So it's going on now three three years. Four years. I think it's going almost for four years. Yeah. Four years now. I think February be four years. I think I think we this is one of the first things we worked on after I after I joined Cloudflare. And yeah, so we still see Russian strikes on Ukrainian infrastructure often times knocking out or targeting power infrastructure and that will ultimately result in drops in traffic that we we can see in certain regions. And that happened initially earlier this year as well already in January. Yeah, we covered one of those like last week. These strikes are ongoing. Car cave more recently. Yes, that was the one this earlier this But those are frequent and it it hasn't stopped actually. I think in the past few months it has increased. There was a time that it stopped a bit and now it has been increasing again. And it reminds us that the war is still ongoing and with a very big impact in terms of the population when it's really cold there both in power and in the internet of course and weather always the case as in in Portugal uh more recently in Portugal yesterday. Yeah. So you know hurricanes and typhoons and and you know all the really severe weather knocks out power it knocks out infrastructure. So you know the the challenge I think and we've talked about this is uh in many cases uh our in many cases for other types of disruptions the traffic graphs are pretty clear and you can say okay this is when it when when the problem happened and this is when they fixed the problem here's when the cable got restored or here's when whatever. The challenge with with things like storm related damage is that it's hard to really pinpoint a this is when it was definitively resolved. So I think we you know we you and I both tend to use our our sort of best judgment in terms of saying okay traffic appears to have recovered to a point where the pattern is more regular um and recover to such a point where it appears that reasonable number of of there's likely a reasonable number of subscribers that are that are able to use the services again. makes sense. Specifically in the in the weather part for example in the last quarter as we we're mentioning it was Jamaica that was impacted for several days Sri Lanka and Indonesia and that was a cyclone sire late November causing floods and landslides multiple day impact as well there in several regions and then we we also had some what is called known or unspecific technical problems. I'm always I always try to figure out how do I title that section. There there are definitely outages and disruptions that we see where it's it's very clear that there's something going on. In some cases the provider will acknowledge that which is great and helpful and they'll say yep yep you know we like like here you know they said telephone SMS and data services are experiencing problems. It's like yeah no kidding. But they didn't say they never say, "Oh, it's because we screwed up DNS or oh, it's because we pushed it back in fig or there was a cable cut or whatever." So, in in some of these cases, they say, you know, yep, there was a problem. You know, in other cases, they will post a follow-up and say, you know, yeah, there was a DNS issue or, you know, um fire at a data center um you know, misconfiguration, you know, whatever it was. So, I think Vodafone uh was one of those where they I believe they had acknowledged this. Oh, no. I take that back. I take that back. So they didn't acknowledge it. I think they they said they didn't provide any information on their social media. Um and then their network status checker was also unavailable which made it hard to go look and see like okay did they add any information there? This was in the UK. This what Vodafone UK. This is Vodafone UK a few hours not multiple days which is usually it does not is not multiple days on countries that are more dependent on the internet like the UK for example. And technical problems are usually fairly quickly resolved. you know, you back out that bad config, you get that DNS server back online or, you know, whatever the case may be. And we saw the announced IP address space dropping for the autonomous system for the phone UK specifically. Usually that's a indicators that they may have screwed up a routing announcement and taken themselves offline unintentionally. Um, that's always an indicator we can look at to see do we have a sense of what's going on there. For sure. Italy also had one fast web in this situation. Yeah, fast. So fast acknowledged the problem. They said, "Yep, it's impacting wired customers." But what one of the interesting things we saw here was, I believe, if you scroll up a little bit more that we saw a spike in Oh, no, we didn't. Sorry. Oh, I'm sorry. Sorry. So, FO actually did say uh that a DNS resolution problem caused the issues. So, it caused the drop in traffic. So in this particular case, what likely happened is that the resolver was experiencing issues and that means that a user or subscriber couldn't resolve the host name for a property served by Cloudflare. So that's why we see the drop in traffic that we do. Yeah. In this case was DNS. Yep. Was it BGP? Was it DNS? We were never Right. And there are some cases where you have, you know, what we've seen the provider says, "Oh, hey, there was a DNS issue." and we can see a uh a corresponding jump in in traffic to to quad one. So you know say it basically tells us hey there's a DNS issue and that users shifted their resolver to use to use cloud flares resolver uh and that sort of got them back online. That's our public DNS resolver what we call quad quad one there. There was also one in Benin specifically. These are common also in Israel cellcom specifically and partner communications. Anything you want to mention? So the the the prior one the one you scroll past in Israel uh said you there was a a publish report that said it may have been a DNS failure. Um so similar to what we saw uh in in fast web. Uh and then at partner communications in Israel u I think that was the one where we saw a uh a spike in queries to quad one. Um so if we scroll up a little bit more you'll see the traffic drop but then you see a spike in traffic to quad one at that same time. Um which could indicate potentially that either users are would indicate two things I guess. one is is that you know users are are shifting to quad one because their local resolver is broken or that there's some issue going on where applications can't reach or or or can't resolve a required host name and they just kind of keep retrying. So we see that occasionally in sort of the what do you call it the the sort of the unidirectional shutdowns. So sometimes queries can reach us but the the resulting traffic can't get back into the country or can't get back into a network. So what we'll see in many cases is is a spike in DNS traffic as like social media apps and things like that try to continually try to retry because they're they're queries are getting out but the responses are not getting back. Makes sense. And uh of course cloud platforms and there was the AWS outage highlighting um some of the stuff that's available on our new cloud observatory page that we have on radar uh that we launched in um in October. And here what we're doing is is leveraging the connections we're making back to these cloud providers as customer origins. So you you may host your origin infrastructure on AWS or Google or whatever. Uh and then when we have to go back and and make a request to that to that origin server, there's connectivity metrics that we can we can measure. So things like success rates and error codes. So if we drill down into like one, you know, you can look at like AWS US East1 or something and just click on click on one of the regions there. Yep. So or or no it's fine. So you know no observed outages over the last four weeks here but what we do see so here's pretty consistent traffic levels. Uh so that's one indicator that something went went crazy if if the traffic level shift pretty consistently low error rates. So there is I guess a little bit spikiness that we've seen. Not totally clear why but again those error rates are still fairly low. And then looking at the connection metrics. So in terms of you know the number of connection failures TCP round trip time is very consistent as is the handshake durations and then the the response time or receive duration as well. So that's you know we're making a request to that server in that region and how long does it take for us to start getting headers back. So you know when any of these metrics begin to have anomalous values we have internal internal alerts set up and we can go sort of investigate you know what metrics are shifting what what if anything are we seeing on the status pages for these providers. So, so trying to correlate our observations of problems with reported problems from the providers. So, and we've seen in previous outages as is mentioned here how the error errors. Yeah. So, you can see the error rate grows there and then at the same time uh we can see the connection failures to try to reach that region went up. So, you're starting to be able to correlate a lot of these metrics to say okay something is definitely wrong because we're seeing these metrics change in a way that indicates there's a problem. Exactly. This was in October of last year specifically. Yeah. And the metrics are going up in this case. Yes. In this case, we're basically saying, you know, it's it's taking longer to try to connect to the systems in this region which is a good metric to show uh and share. There was also the M Microsoft Azure impact here that we also a week later. Yeah. The the I mean the impact so we don't have you know absolute numbers there. Uh but you can see that the um the shifts in metric were not quite as se arguably not quite as severe uh as what we saw with AWS of course and CLER also impacted here. Yep. The couple of issues we had in November and December. Exactly. We've covered those in associated blog posts. Exactly. Makes sense. David, regarding the the Q4 was not uh a lot of events, but there there was a bit of different events in a sense. What would be the key takeaway? I think that you know in in the face of all of it, the internet for the most part still remains resilient. Varying levels of resiliency in the different countries. I think that resiliency frankly also depends on the the level the the I'll say the maturity of the infrastructure. So in some cases uh you know if if there's a lot of um points of centralization then that becomes more problematic things become riskier if there's you know a lot of exposed infrastructure so a lot of you know uh non- buried power lines or or things like that you know certainly many countries and even parts of every country I think some of the some of the infrastructure is is just up on poles and out in the open some of it is buried so you I think the more exposed the infrastructure is the more potential damage that can occur when there's a storm or or you know other sorts of severe weather. Uh and that places internet connectivity at greater risk. Makes sense. Before we go, I don't resist and and I'll share our cloth radar year in review for 2025 in this case showing the internet outages. Of course, those that are hearing in the podcast format, they won't see the image here, the the world map here showing the outages of the year in a sense, the main outages or disruptions. Those are the major ones we saw through early December last year. Exactly. And but it's quite interesting to see that uh those are still ongoing and occurring, right? Where where the hot spots are, you know, where which which regions which countries have a number that's greater than one. Yeah, you could definitely see where is more frequent for sure. Where the internet is a little bit more um impacted. Uh so maybe next year we should look at coloring that actually a little bit more. you know, using a color scale that reflects the uh the severity of the volume, the magnitude of the of the number. Makes sense. Makes sense. Like a heat map where it's darker if there was more disruptions in a sense. And the other the other challenge with this actually as well is is that we're starting to have some providers that are supra national. So providers like Starlink. So the question now becomes how do we capture an issue on a provider like Starlink which has effectively global services. So we can we can assign that to the US because that's where they're based. But the reality of it is that they're they're affecting, you know, all the all the hundreds of countries that they offer services and of course they're global, so it makes sense. It was great, David. Thank you so much. And that's a wrap. Great. Always happy to be here.