Aligns with what I'm seeing. Worse for those services, such as KSAT, is companies like Amazon and Microsoft are really beefing up their networks. Microsoft basically offers a cargo container it can put anywhere, hook it into fast networks, and away you go. It talked about that offering in a co-announcement with Gwynne Shotwell (in 2021, I think).
If there were one thing that truly concerned the military about using commercial networks, it's the possibility of latency. Now, I'd argue that a constellation like Starlink would help significantly against that, but the DoD is very conservative in its approach and I could see them using both, falling back to the legacy systems when something like Starlink boofs it.
Long time listener -- first time caller. Thanks Joe for your constructively provocative thought service. I certainly appreciate that moving mindsets (esp government) takes a very long time. Relative to this recent thread of yours, I’ve been advocating replacing the term “direct downlink” with “direct access” for a couple of reasons: 1) it is in fact what the user actually wants -- and 2) it avoids assigning the solution to the problem. I do agree that many in govt will want/need to add “assured” to that phrase (thus, “assured direct access”) -- but I’d argue that has been (and will be) a continuum. And, I’d humbly observe, that if a military unit in extremis needs the data to survive a real-time threat, they’ll take the access and worry about the assuredness later. Onward.
The potential is (obviously) awesome, but practically comes down to geometry issues, doesn't it? Line of sight between the antenna face of the relay sat and the EO sat will depend on the EO sat being below the relay sat, and having an uplink antenna pointing behind it. And the relative distance between those altitudes will affect "access" time between a given EO sat and a particular relay sat (can you seamlessly move an upload in progress from one relay sat to another?). Obviously not impossible technical challenges, but will require changes to the design of EO sats in the future (I would think anyways, I don't work in the industry).
Also wonder why the concept hasn't caught on more with using commercial SATCOM out in geo? Available bandwidth, flexibility, costs? Apart from Airbus advertising it, I haven't seen any other providers using it.
Always appreciate the commentary! Good luck to you and the Umbra team in the new year!
Jan 3, 2023·edited Jan 3, 2023Liked by Joe Morrison
Perhaps instead of surge pricing, a better analogy would be the bidding model that Google Ads uses. If Small Co. has an Ads account and is trying to show their ad when users meeting certain criteria are searching for a particular keyword, Small Co. can specify a maximum they are willing to pay for that ad impression. If the bidding from other advertisers gets above that that preset limit, Small Co. sits that one out and the algorithm will wait for the next time a user meeting the criteria searches on that keyword.
It's still not a perfect analogy but the model is structured around addressing demand whereas surge pricing - as you pointed out in the article - is also designed to draw more drivers (supply) into the ecosystem.
How is Part III coming? ... seems, given satellite optical mesh networks, Processing just becomes a node activity in the network and can be done anywhere ... in space or on the ground ... just a matter of the number of hops and the latency requirements ... hops with low latency and high bandwidth.
Okay, this is a "pre-comment" in anticipation of Part III. :)
- how likely is location based surge demand? Low IMHO
- how more likely are on board capacity surge constraints (be it short term storage, compute demands, offload link availability - down or across)? Not as low IMHO
- how far will unit cost of space storage and edge compute fall and how does TCO compare (inc lifetime ESG/NZ factors) with ground station?
- could additive printing in space (IOSM) lead to automated data centre build and expansion in space (just send up some bags of appropriate material and a design pattern) in times not too distant?
- if most value of location observation lies in detecting change and then notifying/acting in response to that change, its the fact (or possibility) of that change that has value not the data itself so the lower the latency the higher the value - does that give the advantage to space based store and compute?
Aligns with what I'm seeing. Worse for those services, such as KSAT, is companies like Amazon and Microsoft are really beefing up their networks. Microsoft basically offers a cargo container it can put anywhere, hook it into fast networks, and away you go. It talked about that offering in a co-announcement with Gwynne Shotwell (in 2021, I think).
If there were one thing that truly concerned the military about using commercial networks, it's the possibility of latency. Now, I'd argue that a constellation like Starlink would help significantly against that, but the DoD is very conservative in its approach and I could see them using both, falling back to the legacy systems when something like Starlink boofs it.
Long time listener -- first time caller. Thanks Joe for your constructively provocative thought service. I certainly appreciate that moving mindsets (esp government) takes a very long time. Relative to this recent thread of yours, I’ve been advocating replacing the term “direct downlink” with “direct access” for a couple of reasons: 1) it is in fact what the user actually wants -- and 2) it avoids assigning the solution to the problem. I do agree that many in govt will want/need to add “assured” to that phrase (thus, “assured direct access”) -- but I’d argue that has been (and will be) a continuum. And, I’d humbly observe, that if a military unit in extremis needs the data to survive a real-time threat, they’ll take the access and worry about the assuredness later. Onward.
Great article. Thanks, Joe.
The potential is (obviously) awesome, but practically comes down to geometry issues, doesn't it? Line of sight between the antenna face of the relay sat and the EO sat will depend on the EO sat being below the relay sat, and having an uplink antenna pointing behind it. And the relative distance between those altitudes will affect "access" time between a given EO sat and a particular relay sat (can you seamlessly move an upload in progress from one relay sat to another?). Obviously not impossible technical challenges, but will require changes to the design of EO sats in the future (I would think anyways, I don't work in the industry).
Also wonder why the concept hasn't caught on more with using commercial SATCOM out in geo? Available bandwidth, flexibility, costs? Apart from Airbus advertising it, I haven't seen any other providers using it.
Always appreciate the commentary! Good luck to you and the Umbra team in the new year!
Perhaps instead of surge pricing, a better analogy would be the bidding model that Google Ads uses. If Small Co. has an Ads account and is trying to show their ad when users meeting certain criteria are searching for a particular keyword, Small Co. can specify a maximum they are willing to pay for that ad impression. If the bidding from other advertisers gets above that that preset limit, Small Co. sits that one out and the algorithm will wait for the next time a user meeting the criteria searches on that keyword.
It's still not a perfect analogy but the model is structured around addressing demand whereas surge pricing - as you pointed out in the article - is also designed to draw more drivers (supply) into the ecosystem.
How is Part III coming? ... seems, given satellite optical mesh networks, Processing just becomes a node activity in the network and can be done anywhere ... in space or on the ground ... just a matter of the number of hops and the latency requirements ... hops with low latency and high bandwidth.
Okay, this is a "pre-comment" in anticipation of Part III. :)
Pt II worth waiting for, roll on Pt III ;-)
Some thoughts:
- how likely is location based surge demand? Low IMHO
- how more likely are on board capacity surge constraints (be it short term storage, compute demands, offload link availability - down or across)? Not as low IMHO
- how far will unit cost of space storage and edge compute fall and how does TCO compare (inc lifetime ESG/NZ factors) with ground station?
- could additive printing in space (IOSM) lead to automated data centre build and expansion in space (just send up some bags of appropriate material and a design pattern) in times not too distant?
- if most value of location observation lies in detecting change and then notifying/acting in response to that change, its the fact (or possibility) of that change that has value not the data itself so the lower the latency the higher the value - does that give the advantage to space based store and compute?
As I say, roll on Pt III......
Great article. Thanks. (New subscriber ... looking forward to reading past articles.)