Posts by ibby

Amazon to pour £8B into UK datacenters through to 2028

Re: Pouring

My guess... 13,800 during construction and around 200 in facilities maintenance, engineering, and telecoms.

Astronomers back call for review of bonkers rule that means satellite swarms fly without environment checks

Re: Consumer thinking

>> Because the really slow part is the up/downlink

Actually no. While it is true that the latency to a geo-stationary orbit is high "Satellite technologies inherently experience longer latencies since packets must travel approximately 44,500 miles from an earth station to the satellite and back. Therefore, the median latencies of satellite-based broadband services are much higher, at 594 ms to 624 ms" [1]. The latency to/from LEO is specified (by Starlink) in the "25-60 ms" [2] range and reported/measured in the "31ms to 94ms" range in 2020 [3] and is currently running at a 32.41ms global average per the independent Starlink Status [4].

For the Starlink satellite connections these measured latencies include both the transmission to orbit and the return plus the terrestrial connection from the downlink station to the target and they are on standard service connections.

And I'm not stating that any LEO constellation can do this today because it requires satellite to satellite (laser) comms and routing in space which are not universally deployed (but are likely a part of the proposed Starshield constellation and next generation Starlink).

Getting out the calculator... The flight distance between the nearest airports New-York (LGA) and London (LHR) is 3,457.54 mi (5,564.38 km) [5] and the refractive index of glass is approximately 1.5 leading to a speed of light in glass of approximately 2x10^8 m/s vs. speed in a vacuum of 3x10^8 (1/3 slower in glass) [6]. The orbital altitude of the Starlink LEO constellation is " at altitudes of 525, 530, and 535 km" [7] so we'll add in an additional 2x535km for the up/down link.

Minimum (one way) transmission time over fiber is: 5,564.38 km / 200,000 km/s = 0.028s = 28ms

Minimum (one way) transmission time over LEO is: 6,634.38 km / 300,000 km/s = 0.022s = 22ms

So while it's not possible today because the technology deployed does not support it there is a 6ms theoretical margin for the LEO solution on a London to New York connection. Neither fiber nor satellite delivers on these theoretical transmission times today but the opportunity is there.

[1] https://docs.fcc.gov/public/attachments/DOC-355405A1.pdf?cl_system=mapi&cl_system_id=&clreqid=&kbid=122820

[2] https://www.starlink.com/legal/documents/DOC-1400-28829-70

[3] https://arstechnica.com/information-technology/2020/08/spacex-starlink-beta-tests-show-speeds-up-to-60mbps-latency-as-low-as-31ms/

[4] https://starlinkstatus.space/

[5] https://www.distance.to/New-York/London

[6] https://www.quora.com/What-is-precisely-the-speed-of-light-in-fiber-optics

[7] https://stlksat.com/wp-content/uploads/2024/08/FCC-22-91A1.pdf

The world is bigger than the US of A

The FCC is a uniquely American institution, it's a national regulator not the worlds LEO regulator. The closest thing to that for comms satellites is the ITU.

https://www.itu.int/hub/2023/01/satellite-regulation-leo-geo-wrs/

Consumer thinking

The consumer business is a way to generate a bit of cashflow, the real money here is going to be in arbitrage. Over oceanic distances (Atlantic or Pacific) the team that transmits data at the speed of light in a vacuum wins big over the team who is tied to the speed of light in fiber.

Google's Privacy Sandbox more like a privacy mirage, campaigners claim

Not sell, use

Google don't sell your data... that would let others make a buck. They monetise your data by themselves to the maximum extent possible.

Microsoft 'retires' Azure IoT Central in platform rethink

Incorrect

The conclusion presented in the article is incorrect. Microsoft's Kam VedBrat has blogged on this topic, acknowledging the message that was presented and advising that it was presented in error and that their commitment remains and signposting where Microsoft posts valid depreciation notices.

https://techcommunity.microsoft.com/t5/internet-of-things-blog/microsoft-s-commitment-to-azure-iot/ba-p/4059725

"There was a recent erroneous system message on Feb 14th regarding the deprecation of Azure IoT Central. The error message stated that Azure IoT Central will be deprecated on March 31st, 2027 and starting April 1, 2024, you won’t be able to create new application resources. This message is not accurate and was presented in error.

Microsoft does not communicate product retirements using system messages. When we do announce Azure product retirements, we follow our standard Azure service notification process including a notification period of 3-years before discontinuing support. We understand the importance of product retirement information for our customers' planning and operations..."