Peter Murray has a post this morning that updates an old conversation he and I had, one that Cliff Lynch also played an indirect part–all riffing off the old note,
When you think you have a really zippy network connection, someone will (should?) bring up an old internet adageL2 which says “Never underestimate the bandwidth of a station wagon full of tapes.”
…which, more recently had entailed versions such as “a truck full of CDs” or, what started this all, “a 747 full of Blu-ray Discs.”
Go read the post. I’ll wait.
In the spirit of scientific investigation (which you can translate as “Because I really should be doing the indexing for the new Cites & Insights, and indexing is really boring…”), I decided to check out a couple of things–e.g.,
- Would 2TB internal hard disks provide even greater bandwidth?
- Would cargo weight or bulk be the limiting factor?
- Which provides greater bandwidth, a 747 full of double-density Blu-Ray discs or a 747 full of 2TB internal hard disks–and what is that capacity (from New York to LA)?
I also changed one thing: Realistically, even double-DVD slimpacks aren’t the way you’d ship all this stuff. You’d use 100 disc spindles, which result in less packaging overhead.
Here’s what I found
Table
| Cargo capacity (cubic meters) | 764 |
| Cargo capacity (kg) | 123,656 |
| Volume of 100 BD spindle (cm) | 0.00347 |
| Weight of 100BD spindle (kg) | 1.316 |
| Max spindles (volume) | 220,173 |
| Max spindles (weight) | 93,964 |
| Data capacity at 40Tb/spindle | 3,758,541 |
| Bandwidth JFK-LAX, Gb/sec | 232,009 |
| Volume of 10 2TB HD (cm) | 0.00390 |
| Weight of 10 2TB HD (kg) | 7.50 |
| Max 10packs (volume) | 195,998 |
| Max 10packs (weight) | 16,487 |
| Data capacity at 160Tb/pack | 2,637,995 |
| Bandwidth JFK-LAX, Gb/s | 162,839 |
Notes
I checked Boeing’s website for the maximum payload capacity of a Boeing 747 freighter (see Peter’s link, but go to other sublinks as needed). I did real-world measurements for the size and weight of a 100-disc spindle and used Western Digital’s own specs for their Caviar Black 2TB internal hard drive–and, to simplify calculations, I assumed “10packs” of the discs, wrapped 10 high in plastic wrap. (I assume plastic wrap throughout rather than boxes, again to simplify things.) The bandwidth calculations assume the 16,200 seconds in Peter’s post.
To explicate what’s here:
- A spindle of 100 Blu-ray discs (total data capacity 5TB or 40Tb) occupies 0.00347 cubic meters (basically, 7.5×5.5×5.5 inches or 177.8×139.7×139.7 millimeters) and weighs 1.316 kilograms (2.9lb.) You could fit 220,173 spindles (in other words, just over 22 million discs) in the 747 freighter–but the plane couldn’t take off. By weight, it could hold 93,964 spindles (just under 9.4 million discs)–so the actual data capacity would be 3,758,541 Terabits, for a bandwidth of 232,009 Gb/s–just a little higher than Peter’s numbers, because spindles add so much less bulk than individual packages.
- A stack of 2TB hard drives 10 high (total data capacity 20TB or 160Tb) occupies 0.003898 cubic meters (261 millimeters high, 147 millimeters wide, 101.6 millimeters deep) and weighs 7.5 kilograms. That’s the killer: While you could fit almost 1.96 million drives into the plane, you could only take off with 166,487 drives (16,487 tenpacks)–so the actual data capacity would be 2,637,995 Terabits for a bandwidth of 162,839 Gb/s.
Both are, to be sure, three orders of magnitude greater than the fastest reported network transmission. I was a little surprised to find that Blu-ray discs offered more bandwidth than hard disks–because a spindle of 100 Blu-ray discs with 5TB total capacity weighs less than two 2TB hard disks.
Another little table:
| BD | HD | |
| Capacity (per cubic meter) | 1441TB | 5131TB |
| Weight (per cubic meter) | 379kg | 1924kg |
Significance and omitted elements
- None…except that the proverbial station wagon full of tapes still has some, erm, legs.
- Many–some of them discussed in the original post and comments.
Now for that indexing…
however, LATENCY is teh sux.
this is a real-time world, old man! get with the frikken’ program!
(insults were meant as a joke. i’m older than thou!)
Thanks for picking up the challenge to take this to the next level, Walt. I discounted Tyvek sleeves and spindles because I was concerned about the potential for damage to the discs. After all, the “cost” of retransmission (mostly in time) is so great when your transport layer is a cargo aircraft. But, then again, if those techniques are good enough to ship new discs, they are probably good enough to ship written discs. Geez — to describe it like this makes one think we’re actually taking this scheme seriously.
Very interesting to see that the bounding variable is weight and not volume!
Peter: You’re welcome. I’d discount Tyvek sleeves too, probably–but spindles with covers (which is what I was measuring, including weight) are probably at least as safe as slimpaks–less breakage-prone and with discs firmly locked in place.
marcbl: You’re over 64? Good to have some older readers. And, actually, you might want to click on the first link in the post again and read the comments–specifically Eric Lease Morgan’s comment about Google and FedExing hard discs as a cheap, fast way to move large quantities of data. Sure, if you needed, say, one gigabyte the net is the way to go–but if what you need is, say, 500 Terabytes, shipping objects around starts to look pretty attractive.
(One reason why the U.S. probably isn’t going to have universal on-demand high-def “any movie or tv program you want, any time” real soon: Not only is most U.S. broadband far too slow for HD, the backbones don’t have the capacity for tens of millions of distinct 30mb/s streams. For certain purposes, broadcast and physical carriers still make lots of sense.)