Zoom Logo

Secure Data Storage - WG - Shared screen with gallery view
Eddie
14:32
I still need to sign it, but I reached out via email and haven’t heard back. eddie@bloom.co is how I emailed.
Manu
15:20
Welcome Jackson and Phillip! :)
Phillip Long
15:32
Thanks Manu!
Dmitri Zagidulin
15:39
agenda document: https://hackmd.io/ckcR0vDYRkucm3eISzB-FA
jackson
15:45
Thanks!
Orie Steele (Transmute)
17:38
See https://github.com/decentralized-identity/secure-data-store/issues/43
Manu
26:06
haha! Pigeons.
Manu
30:58
huzzah!
Manu
31:05
+1 for ipld://....
Dave Longley
31:09
+1
Orie Steele (Transmute)
32:35
https://uniresolver.io/#did:ipid:QmYA7p467t4BGgBL4NmyHtsXMoPrYH9b3kSG6dbgFYskJm
Orie Steele (Transmute)
34:29
https://www.iana.org/assignments/cose/cose.xhtml
Orie Steele (Transmute)
34:42
https://www.iana.org/assignments/cbor-tags/cbor-tags.xhtml
Dave Longley
35:35
Orie: I get no result at that first link :(
Orie Steele (Transmute)
35:46
Yep, its been like that for a while
Orie Steele (Transmute)
35:55
But that its the example in the universal resolver
Dave Longley
36:06
ok, perhaps related to that comment Jonathan made about needing to get an update to Markus.
Dmitri Zagidulin
36:15
q+
Orie Steele (Transmute)
36:16
Yes I think so
Dmitri Zagidulin
37:34
in-progress Layers diagram that might be helpful for people: https://docs.google.com/presentation/d/1QEHSs4XJ05yQl2mvpiqbM80-MySxlVI9cNDLPq_XkkY/edit#slide=id.g8a2289ca98_0_0
Dmitri Zagidulin
38:09
q+
Manu
38:28
q+ to note a couple of places I think IPLD could fit...
Orie Steele (Transmute)
38:44
q+
Manu
39:23
q- later (after Orie)
Dave Longley
42:17
+1 to Orie's questions, how do we get stable mutable identifiers in here and have an authority model around data access?
Dmitri Zagidulin
42:28
+1, had that same question
Adrian Gropper
42:53
I’m so lost….
Dmitri Zagidulin
44:20
https://docs.ipfs.io/concepts/ipns/
Dave Longley
44:58
Sounds like you need a keypair for every stable link and ... yes, what Orie is saying, no portability.
Wayne Chang
45:27
when didcomm
Dmitri Zagidulin
45:35
lol
Orie Steele (Transmute)
46:54
Daniel Buchner how much. Of this is the “Hubs dream interface” ?
Manu
47:02
q+
Orie Steele (Transmute)
47:17
It sounds close…. But the portability issues see problematic
Adrian Gropper
47:53
q+ to understand even one use case
Orie Steele (Transmute)
49:14
I think IPNS + JWE will work got documents as is
Orie Steele (Transmute)
49:39
Thinking of the IPNS server the EDV Server.
Dave Longley
50:06
It seems like you need one key per stable mutable document ID -- and that you'd need to share some kind of access to use that key to share write access.
Orie Steele (Transmute)
51:16
Yep… which means when that key is blown…. So are you
Orie Steele (Transmute)
51:59
Portable mutable identifiers… seem like guide would be better.
Wayne Chang
52:13
+q
Wayne Chang
52:20
q+ even
Dave Longley
52:33
My biggest concern is that all of the benefits of these systems are that you don't need authority to move data around -- but we *need* that authority for EDVs... which throws a wrench into the works; we can't take advantage of the beenfits.
Dave Longley
52:36
benefits*
Manu
53:08
q+ to highlight use case, and dangers of IPFS (public mode), not private mode, private mode may be great.
Manu
53:36
Not solely your problem, Adrian -- we're grappling w/ finding right use case.
Dave Longley
54:04
Note that if "private mode" is within a single provider -- we're not really gaining much benefit, it's just an implementation detail.
Orie Steele (Transmute)
54:21
Yes, thats like a mongo db cluster
Dave Longley
54:39
Yeah :/
Dave Longley
56:05
q+
Adrian Gropper
56:17
Datashards vs. IPFS in replication?
Orie Steele (Transmute)
56:20
The public key addressing scheme seems to be the crux of the issue…
Dave Longley
56:41
q+ to talk about replication within a storage provider vs. across storage providers
Orie Steele (Transmute)
56:49
3DES FTW
cwebber
56:55
just to be sure: there's no official standard for encrypted content on ipfs yet right anyway... unless I'm wrong?
Wayne Chang
57:14
+1...there’s good reason why PHI is still considered PHI even when encrypted
cwebber
57:18
it's a per-application decision on how to encrypt content, correct?
Dmitri Zagidulin
57:44
@chris - feel free to get on the q to ask. But I think the encryption is an in-progress spec by jonathan
jackson
57:53
+1 on no encrypted content on public networks
Orie Steele (Transmute)
58:26
Yes, in order to escape storage… it needs to handle public network access.. which touches authorization… and thats where things become troublesome
Adrian Gropper
58:31
Replication is not what I might call a use-case. It’s a feature.
Dmitri Zagidulin
59:25
+1 adrian
Adrian Gropper
01:02:15
+1 to @Dave and replication controlled by the PDP.
Dave Longley
01:06:36
IPLD seems like it could be useful as a manifest format for chunks.
Manu
01:06:45
yes, +1 to that
Manu
01:07:12
especially if there is a URL format for IPLD, which I think was shown?
Orie Steele (Transmute)
01:07:45
Yes, I feel IPLD could be great for storage layer, and in the case of replicating in a data center… it would also be useful.
Orie Steele (Transmute)
01:08:14
It starts to be hard when you start thinking about migrating from Apple to Amazon, etc...
Juan Caballero
01:09:06
next week as well?
Juan Caballero
01:09:23
4pm my time, 7am pst
Juan Caballero
01:09:25
wed
Juan Caballero
01:09:29
on zoom