Hacker News Reader

Comment by antihero | Hacker News Reader

> Boeing 747 has a bandwidth of 1 TBps.

Hmm, not sure about those numbers.

Say we limit it to this trip, at 10 hours.

A Boeing 747 can carry ~30,000 kg (http://www.boeing.com/commercial/747family/pf/pf_facts.html) You can get get 3TB 3.5' drives now, and they weight about 1kg (http://www.amazon.co.uk/WD-Caviar-Green-WD30EZRX-SATA-600/dp...), so 30,0003TB is 90,000TB in 10 hours.

10 hours is 36000 seconds, so that means that the speed of the data on a 10 hour flight is actually 90000/36000 = 2.5TB/sec.

That said, if we were to account for the next generation jump (http://www.techspot.com/news/47860-seagate-60tb-hdds-possibl...) so 60TB for about a kilo, we'd get 6030000 / 36000 = 50TB a second.

Obviously, the longer the flight, the more inefficient this transfer becomes.

View on HN

> A Boeing 747 can carry ~30,000 kg

More like four times that: the 747-400 has a payload of 112 metric tons (248,600 pounds, for an MTOW of 412 metric tons).

http://en.wikipedia.org/wiki/Boeing_747-400

If you have the data on 32GB micro SD cards (at about three to the gram), you're at 96TB/kg, or 2.88 exabytes/747. On a ten-hour trip, that's 80TB/s. It wouldn't be at all practical to ship 'em loose, and finding and using the data you want from the "stream" would be a real SOB, but that's a lot of raw throughput with current, non-exotic tech.

Of course it's only the end-to-end throughput that counts. It might take quite a while to load the data onto all those micro-SDHC cards and then extract it again at the destination, and you have to add time that to the ten-hour flight time. Not to mention the health-care bill for driving your sysadmin insane.

My point, really, was that even with inefficient and inappropriate packaging, the available raw transfer rate is already huge. It's just a matter of packaging what already exists into "data cargo containers" sized and priced appropriately.

You could use just the chips from micro SDHC cards which weigh 0.5g and have 64GB capacity. So you could fit 128 TB into 1 kg

An other, middle-of-the-road (and probably cheaper) option is LTO-5 tapes at 1.5TB for 270g (5.55 TB/kg).

(and according to wikipedia a µSD card is 0.25g)

>Obviously, the longer the flight, the more inefficient this transfer becomes.

However, it's a purely linear degradation with a hard upper limit.

http://wdc.com/en/products/products.aspx?id=120#tab3 says 3TB weighs 0,74 kg. So that would make 40.540 drives -> 121,620TB in 10 hours. But I'm not sure they would all fit due to their volume relative to airplane's.

You should be measuring MicroSD cards, not hard drives.

I'll modify the post to make that point. You are correct that the real bandwidth of a single 747 is more like 220 Mbps.

In either case, it's an interesting exercise but missing some key calculations...

When I transfer data from machine to machine, the data is immediately "usable" on the receiving machine.

When transferring data by physical means, the data needs to be brought to the final point of consumption, and possibly loaded into the host machine if it is not in some form that could be immediately connected/mounted.

So, the real comparison would be transferring data from the host machine to some form of removable media, taking that media to the airplane (early enough to meet all the pre-clearance stuff), waiting for the plane to takeoff, fly, land and then handoff the physical media at the other end.

In many cases, I'm sure the plane is still faster, but if we're talking about its theoretical bandwidth we need to look at the entire "trip" of the packets/data, not just one segment of it.

Unless you take the rackspace crates off the 747 and hook them up at the destination end. But compare that to ordering blank disks, having them delivered, configuring them and then copying network data onto them.

I guess that you were right and you do not need to change anything. 220 Mbps is the throughput and not the bandwidth.

Assume a scenario wherein you change the destination to Mumbai. Your throughput will get reduced to 110 Mbps (assuming a 20 hour flight), while the bandwidth will stay the same as the networking medium (airline) has not changed, right?

Couple of other points -

* Two media with same bandwidth can have different throughput depending on the distance (i.e latency) or packet loss or other reasons. Consider two 100 Mbps broadband lines from SFO to London. One goes via the east coast, while other via Asia. Both will have different throughput; difference being roughly proportional to the difference in distances.

* 220 Mbps would imply that a maximum of 220 Mb data can pass through the medium, which is not the case, since we are loading TBs of data on the plane.

* The window size in case of an airplane would be the size of the data loaded on it i.e the size of the movie collection.

Out of curiosity, what caused the error in the article?

I was wondering if you derived the number via a similar calculation as above (a full 747), it's a direct reference to the size of the hard drive it's carrying, or just a simple typo?

It was an editing error. I originally had a whole paragraph imagining a continuous line of 747s containing 1TB disks and was going to be making an analogy about the connection across the Internet, but it was too complicated. In editing that out I didn't go back to the right number.

Reply on news.ycombinator.com