Posts by JanCeuleers

Alpine Linux 3.18 fixes DNS over TCP issue, now ready for all the internet's problems

The "so what" portion is missing from the article

I'm not an expert, but one reason I'm aware of why DNS over TCP is required is that only over TCP can large DNS replies be handled (such as when looking up TXT records in the context of SPF verifications done by incoming mail servers). Since UDP does not guarantee that segments are transmitted in order, or indeed at all, segmentation of large packets is supported only over TCP. The RFCs say that a DNS query initially made over UDP must be retried over TCP if the reply was too large to fit into a single packet.

Toyota's bungling of customer privacy is becoming a pattern

Seems like basic controls, such as internal audit, were missing.

Should regulators require companies suffering data breaches to achieve ISO27K certification?

Hospitals cancel outpatient appointments as Irish health service struck by ransomware

Nationalised?

Something can only be nationalised if it was previously private.

Where's our data, Google? Chrome 79 update 'a catastrophe' for Android devs with WebView apps

Similar to a supply chain attack

These apps' developers should perhaps view their dependence on 3rd-party-provided functionality as a supply chain attack vector and act accordingly.

I am not saying that these developers are at fault for causing this issue, but it is clear that they have taken a risk by relying on the 3rd-party functionality, which makes that functionality part of their supply chain.

Sure, we made your Wi-Fi routers phone home with telemetry, says Ubiquiti. What of it?

GDPR compliant?

Since GDPR requires "informed consent" from people before their data is processed (gathered, transmitted, analysed, acted upon, ...), whether it is anonymised or not, it is not possible for Ubiquiti's data gathering to be GDPR-compliant without people having given that consent.

Apple asks FCC to let it run mm-wave tests - for backhaul?!

Re: Come to think of it ...

I took a look at that, using Google Maps and the distance/elevation calculator at daftlogic.com, and I still don't think it flies.

The straight-line distance between the Black Mountain and the Milpitas campus is 23km, which at 29GHz corresponds with about 2.8dB of attenuation due to atmospheric absorption (using the attenuation curve shown in the first reference to the EHF article on Wikipedia - note that the curve shown in the article itself corresponds with a much higher elevation - since this case is much nearer sea level curve A in the book applies).

Anyway, 2.8dB seems great, until you take into account that mm waves severely suffer from rain fading, foliage blocking, scattering etc., each of which are impossible to engineer for and therefore render the link highly unreliable).

The fact that at 10km the Cupertino campus is much closer to the Black Mountain doesn't change my opinion.

All of this in addition to the fact that the subject of mm waves and 5G is about the air interface, not backhaul. The "for backhaul?!" quexclamation in the title is clickbait.

The 20-25 degree downtilt invalidates your theory that this is about backhaul. Even accounting for the 217' of elevation differential between Cupertino and Milpitas such an antenna downtilt results in the signal hitting the ground within a few hundred yards from the tower. Moreover for backhaul you'd want a much narrower beamwidth.

Betteridge's law strikes again.

Geo-boffins say 'quake lifted bits of New Zealand by 8 metres, moved at 3km/second

Re: Sandwich filler?

The title says that NZ moved by 8m, whereas the article talks about two metres. Which is it?

TfL to track Tube users in stations by their MAC addresses

What about MAC address randomisation of the kind first introduced by Apple but that has now also been implemented in wpa_supplicant?

'Legacy' Wordpress blog site of The Independent serving malware

Are ad flingers suppliers

Surely advertising brokers are customers of websites on which their adverts appear, not suppliers?

The modest father of SMS, who had much to be modest about

ISDN had user-to-user signalling

An ISDN feature that predates SMS is user-to-user signalling (UUS). It comes in three flavours: UUS1 involves the ability for short messages to be included in the setup and cleardown messages (and the service involves the network transporting these messages end-to-end. UUS2 involves the ability to send such messages during the alerting phase of a call, i.e. while the "phone is ringing" (or the equivalent for a data call). Finally UUS3 involves the ability to exchange text messages during the active phase of a call.

In all cases, as you can see, are UUS messages associated with a call whereas SMS messages are not. But it would have been possible to leverage UUS1 for the purpose of implementing an SMS service (i.e. a call setup message that transports UUS1 and with implicit cleardown.

UUS was not at all widely used or supported by public networks. One potential use case was for signalling interworking between PBXes that used the public ISDN for interconnection (as opposed to leased lines).

Intel has ambitions to turn modems into virtual servers and reinvent broadband

Contrary to industry trend

CPE provided by telecoms service providers (often referred to as home gateways) are felt to already be too complex and too difficult to deploy upgrades to. So the industry trend is to virtualise these by pulling functionality into the service provider cloud, rather than to put more grunt into the CPE so as to also put more functionality and complexity into them.