[SIO-SWAP] Name service and File exhange on SWAP

[Dale Chayes]

A couple of comments on the name service question, and perhaps for all 
aspects of SWAP...

In my opinion, a viable solution should work for:

	- ships that have little or no onboard networking/sysjan talent
	- ships w/ already over-committed onboard support personnel
	- ships that do not have "regular" IP connectivity

I know that this raises the bar, but we have to be realistic about 
making this appealing and scalable (rather than a liability) for a 
significant fraction of the fleet, most of which by ship-count fall in 
the small boat, small over-committed-staff category.

-Dale

On Mar 5, 2004, at 10:47 AM, Steve Foley wrote:

>
>
> Val Schmidt wrote:
>> We could alternatively leave the swap boxes as they are and let 
>> individual
>> ships collect the dns file when they want and serve it up locally 
>> from their
>> SWAP Host. This prevents us from having to mucky with dns on the swap
>> devices and puts the onus on the ship/institution.
>
> Maybe Im not completely understanding the issue (got my head in 
> another network right now), but DNS is really just a way to map names 
> to numbers. If that's all we are looking to do, why not delegate 
> control of a subdomain to each ship, and let them advertise their 
> name/number mapping. For example, Melville could be the DNS master for 
> melville.swap.unols.org (or whatever domain). In each DNS setup, it 
> would be helpful to hand out the name/number mappings for the DNS 
> servers. Yes, this would hammer up each ship's DNS server address by 
> hand, but if SWAP DNS is handled by the SWAP device, dont we know this 
> address already?
>
> The ship is going to know what names and numbers are most 
> appropriate/accurate/applicable to its ship. Pushing the knowledge of 
> what hosts are onboard to an entity off the ship seems counter 
> productive and an invitation for everyone to get out of sync.
>
> However, if we are looking for an even trickier way of advertising 
> dynamic computing services across the network, then we are talking 
> about something more than just name/number mapping, so DNS really isnt 
> the right technology (at least in its current form).
>
>
>> File Transfer:
>> --------------
>> There are a few flavors I've thought about.
>> 1) Nothing automated. You just plop the files you want to share on a 
>> web
>> server and let the peer ship browse for them and download what they 
>> will.
>> Low tech.
>
>> I like option 1 best for low tech ships and institutions and option 3 
>> best
>> for higher tech ones.
>
> During the initial stages of SWAP thought, #1 was the scenario that we 
> had envisioned. Lately, Ive been working with people trying to swap 
> files with shore through our satellite system (much like I suppsoe 
> they would with another ship). Surprisingly or not, I have found that 
> (a) scientists have different equipment and ways of working and (b) a 
> lot of the preferred methods are remarkably low tech. Yes, its true 
> that some scientists will be willing to do whatever they need to for 
> files to be swapped, but I think a lot of scientists will really want 
> to swap files their way. Some bring macs, some PCs, some no machines 
> at all. Some want FTP right to their destination from their laptop, 
> others would rather nfs and copy to a server that will proxy it across 
> the link. Generally, people just like to email files.
>
> The bottom line is that I think file transfer is going to be a moving 
> target that varies with the science party. Id say let's standardize on 
> a really simple method to make sure all SWAP vessels can exchange 
> files, then let the individual scientists decide how they want to do 
> it. Chances are, if they are swapping files with people on the other 
> ship, they already have (or will have to anyway) agree on what gets 
> sent and how it will be received.
>
> Just my $0.02,
> -Steve
>
> _______________________________________________
> SWAP mailing list
> SWAP@sio.ucsd.edu
> http://siomail.ucsd.edu/mailman/listinfo/swap