Skip to content
This repository has been archived by the owner on Apr 29, 2020. It is now read-only.

Open Problem: Mutable Data (Naming, Real-Time, Guarantees) #6

Merged
merged 13 commits into from
Oct 31, 2019
Merged

Open Problem: Mutable Data (Naming, Real-Time, Guarantees) #6

Changes from 2 commits
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
b77e57f
Create MUTABLE_DATA.md
daviddias Sep 10, 2019
6fe11b0
Update MUTABLE_DATA.md
daviddias Sep 11, 2019
03bb262
Update MUTABLE_DATA.md
daviddias Sep 11, 2019
e90a484
Update MUTABLE_DATA.md
daviddias Sep 18, 2019
3f6b2db
Update MUTABLE_DATA.md
daviddias Sep 18, 2019
38a4a7e
Update OPEN_PROBLEMS/MUTABLE_DATA.md
daviddias Sep 26, 2019
9874ca2
Update OPEN_PROBLEMS/MUTABLE_DATA.md
daviddias Sep 26, 2019
302c13a
Update OPEN_PROBLEMS/MUTABLE_DATA.md
daviddias Sep 26, 2019
cfe7fad
Update OPEN_PROBLEMS/MUTABLE_DATA.md
daviddias Sep 26, 2019
80a48d7
Update OPEN_PROBLEMS/MUTABLE_DATA.md
daviddias Sep 26, 2019
0e887da
apply Jorge review for a second interation
daviddias Oct 30, 2019
224b7a4
Update MUTABLE_DATA.md
daviddias Oct 30, 2019
c4fa5cd
update the solution bit
daviddias Oct 31, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
135 changes: 135 additions & 0 deletions OPEN_PROBLEMS/MUTABLE_DATA.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,135 @@
# Mutable Data (Naming, Real-Time, Guarantees)

## Description

Mutable Data in a Permanent Datastore? WAT?! If this is your reaction, don't despair, we've been there too. However, the good news is that achieving mutability over immutable datastores is not a new idea, in fact, if you are checking this document through Git/GitHub, you are experiencing just that.
daviddias marked this conversation as resolved.
Show resolved Hide resolved

The simple trick is to use pointers. While each data struct will have its own reference (in the case of IPFS, they will have a CID), mutability is achieved by updating the pointer to point to a new version of that data structure, just like when you update your master branch (pointer) to your latest commit (immutable reference) on your Git projects.

Having Mutable Data is essential to create a whole new set of applications for the Web 3.0. These applications will have different types of requirements, from millions of records being update every second, to millions of users interacting and mutating a data source.
daviddias marked this conversation as resolved.
Show resolved Hide resolved

Creating a sound solution to achieve Mutable Data for the dWeb requires considering the three parts of the problem:

- **Update Propagation** - As noted, diffent applications will have different requirementsm, these can go from real-time delivery, to delivery at least/most once, confirmation of delivery, conflict resolution, etc.
daviddias marked this conversation as resolved.
Show resolved Hide resolved
- **Update Authentication** - How to trust the parties that are propagating the updates, how to trust that they are not hiding any important update
daviddias marked this conversation as resolved.
Show resolved Hide resolved
- **Interface/API** - This is where things that shape up to feel like a database start to show. From convinence features to powerful APIs that enable a user to query, update and search over multiple records
daviddias marked this conversation as resolved.
Show resolved Hide resolved

All of these problems should be solved to work on a Distributed Network scenario, this means: disconnected, offline first and over unreliable networks.
daviddias marked this conversation as resolved.
Show resolved Hide resolved

## State of the Art

> This survey on the State of the Art is not by any means complete, however, it should provide a good entry point to learn what are the existing work. If you have something that is fundamentally missing, please consider submitting a PR to augment this survey.

### Within the IPFS Ecosystem
> Existing attempts and strategies

##### IPNS
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@hugomrdias @aschmahmann where can I find your latest slides, talks and notes on IPNS with multiple routers?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also include:

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There's also this issue: ipfs/specs#198


- https://docs.ipfs.io/guides/concepts/ipns/

##### dnslink

- https://docs.ipfs.io/guides/concepts/dnslink/

##### libp2p's floodsub/gossipsub

- https://github.com/libp2p/specs/blob/master/pubsub/README.md
- https://github.com/libp2p/specs/blob/master/pubsub/gossipsub/README.md

##### peer-base's CRDTs implementations

- https://github.com/ipfs-shipyard/js-delta-crdts
- https://github.com/ipfs-shipyard/peer-crdt
- https://github.com/ipfs-shipyard/peer-crdt-ipfs

##### orbit-db

- https://orbitdb.org/

##### Textile's data wallet and caffee

- https://docs.textile.io/concepts/the-wallet/
- https://docs.textile.io/concepts/cafes/

### Within the broad Research Ecosystem
> How do people try to solve this problem?

There are multiple decades of Research and Projects to solve this Open Problem. Don't feel disencourged by the sheer amount of articles, it is really high valuable work!
daviddias marked this conversation as resolved.
Show resolved Hide resolved

- [Compreensive PubSub Reading List](https://ipfs.io/ipfs/QmNinWNHd287finciBwbgovkAqEBQKvnys1W26sY8uupc5/pubsub%20reading%20list.pdf)
- [Conflict-free replicated data types](https://scholar.google.pt/citations?view_op=view_citation&hl=en&user=NAUDTpMAAAAJ&citation_for_view=NAUDTpMAAAAJ:M3ejUd6NZC8C)
- [A comprehensive study of Convergent and Commutative Replicated Data Types](http://hal.upmc.fr/inria-00555588/document)
- [Merging OT and CRDT Algorithms](http://dl.acm.org/citation.cfm?id=2596636)
- [Delta State Replicated Data Types](https://arxiv.org/abs/1603.01529)
- [CRDTs: Making δ-CRDTs Delta-Based](http://novasys.di.fct.unl.pt/~alinde/publications/a12-van_der_linde.pdf)
- [Key-CRDT Stores](https://run.unl.pt/bitstream/10362/7802/1/Sousa_2012.pdf)
- [TRVE Data: Placing a bit less trust in the cloud](https://www.cl.cam.ac.uk/research/dtg/trve/)
- [LSEQ: an Adaptive Structure for Sequences in Distributed Collaborative Editing](https://hal.archives-ouvertes.fr/hal-00921633/document)
- [A Conflict-Free Replicated JSON Datatype](https://arxiv.org/pdf/1608.03960.pdf)
- [OpSets: Sequential Specifications for Replicated Datatypes](https://arxiv.org/abs/1805.04263)
- [Snapdoc: Authenticated snapshots with history privacy in peer-to-peer collaborative editing](https://martin.kleppmann.com/papers/snapdoc-pets19.pdf)
- [Access Control for Weakly Consistent Data Stores](http://www.complang.tuwien.ac.at/kps2015/proceedings/KPS_2015_submission_25.pdf)
- [ACGreGate: A Framework for Practical Access Control for Applications using Weakly Consistent Databases](https://arxiv.org/abs/1801.07005)
- [A CRDT Primer Part I: Defanging Order Theory](http://jtfmumm.com/blog/2015/11/17/crdt-primer-1-defanging-order-theory/)
- [A CRDT Primer Part II: Convergent CRDTs](http://jtfmumm.com/blog/2015/11/24/crdt-primer-2-convergent-crdts/)
- Talks
- [RedisConf18: CRDTs and Redis—From Sequential to Concurrent Executions by Carlos Baquero](https://www.youtube.com/watch?v=ZoMIzBM0nf4)
- [QCon London 2018: CRDTs and the Quest for Distributed Consistency by Martin Kleppmann](https://www.infoq.com/presentations/crdt-distributed-consistency)
- ["CRDTs Illustrated" by Arnout Engelen](https://www.youtube.com/watch?v=9xFfOhasiOE)
- [Coding CRDT](https://www.youtube.com/playlist?list=PLzUeAPxtWcqxBXjUelmcm5ORVjEpbUlHH)
- [Dmitry Ivanov & Nami Naserazad - Practical Demystification of CRDT (Lambda Days 2016)](https://www.youtube.com/watch?v=PQzNW8uQ_Y4)
- [ElixirConf 2015 - CRDT: Datatype for the Apocalypse by Alexander Songe](https://www.youtube.com/watch?v=txD1tfyIIvY)
- [GOTO 2016 • Conflict Resolution for Eventual Consistency • Martin Kleppmann](https://www.youtube.com/watch?v=yCcWpzY8dIA)
- [CRDTs in IPFS](https://www.youtube.com/watch?v=2VOF-Z-nLnQ)
- [Journal Club - 2018 06 13 CRDT JSON Datatype, by Gonçalo Pestana](https://www.youtube.com/watch?v=TRvQzwDyVro)
- [CRDT Tutorial for Beginners](https://github.com/ljwagerfield/crdt)
- [Conflict-Free Replicated Data Types (CRDTs), An Offline Camp passion talk](https://medium.com/offline-camp/conflict-free-replicated-data-types-crdts-2c6ae67ab9a4#.duh4g0r9k)
- [CRDT Notes by Paul Frazee](https://github.com/pfrazee/crdt_notes)
- [Towards a unified theory of Operational Transformation and CRDT by Raph Levien](https://medium.com/@raphlinus/towards-a-unified-theory-of-operational-transformation-and-crdt-70485876f72f)
- [A simple approach to building a real-time collaborative text editor](http://digitalfreepen.com/2017/10/06/simple-real-time-collaborative-text-editor.html)
- [Data Laced with History: Causal Trees & Operational CRDTs](http://archagon.net/blog/2018/03/24/data-laced-with-history/)
- http://y-js.org/
- http://swarmdb.net/
- http://gun.js.org/
- http://scuttlebot.io/
- https://github.com/mafintosh/hyperlog
- https://github.com/jboner/akka-crdt

### Known shortcommins of existing solutions
daviddias marked this conversation as resolved.
Show resolved Hide resolved
> What are the limitations on those solutions?

## Solving this Open Problem

### What is the impact

Mutable Data is a fundamental building block for creating applications. Improving the existing solutions and creating new ones that will support a large number of users in a distributed context, will enable the entry of the large growing number of Internet users to be part of the dWeb.
daviddias marked this conversation as resolved.
Show resolved Hide resolved

### What defines a complete solution?
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not necessarily happy with this section. I feel it is not clear enough on how ambitious and challenging solving this Open Problem is.

@jsoares @yiannisbot Would you like to review and make some suggestions? I'm out of creative juice :D

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

TLDR: A complete solution for mutable data (in a p2p context) is pretty unreasonable given that we don't have a one-size-fits-all solution in the centralized world.

I suspect that the reason it's not clear is that there isn't (or has not yet been defined) a one-size fits all answer for what mutable structures should look like or how they should work. Some examples:

  • What form of access control/capabilities even makes in a non-centralized context?
    • Consensus, eventual consistency, causal consistency, etc.
    • Should ACLs be changeable?
    • If cryptography is utilized for ACLs how is it upgraded?
  • What assumptions can people work with?
    • This problem maps really closely with solving distributed identity, but most solutions assume pre-distributed identity... it's a bit of a contradiction
    • If cryptography is utilized which types are ok, which math is sufficient
  • How do people find new mutable data?
    • This might seem unrelated, but given its tie into the identity problem as well as potential forking of mutable data streams it's pretty important

I suspect that step 1 in any "complete solution" is defining the tradeoffs and composability properties of solution categories. This is perhaps a combination of something like ipfs/notes#379 with a more precise version of the Zooko's triangle tradeoffs (it is notoriously underdefined/specified and could probably be more complete if explicitly correlated to the CAP theorem).

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you for the notes @aschmahmann! :)

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Took your notes, restructured a bit and expanded the section more. Thank you!

@yiannisbot @jsoares your feedback is still very welcome. I'll go ahead and call this one done, PRs are always accepted :)

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@yiannisbot @jsoares your feedback is still very welcome. I'll go ahead and call this one done, PRs are always accepted :)

Happy to have a look, but can only do tomorrow.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have a bunch of language fixes but those can wait for @yiannisbot's comments. The version incorporating Adin's comments is already much more detailed but, as acceptance criteria for an RFP, it's still fuzzy.

Saying that being the easy part, coming up with objective criteria probably requires some discussion of system parameters and requirements, which is not critical at this stage but should be done at the RFP drafting stage (see nice example here).

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

as acceptance criteria for an RFP, it's still fuzzy.
..
which is not critical at this stage but should be done at the RFP drafting stage (see nice example here).

Exactly, this docs is not the RFP and more broad in nature

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I suspect that step 1 in any "complete solution" is defining the tradeoffs and composability properties of solution categories. This is perhaps a combination of something like ipfs/notes#379 with a more precise version of the Zooko's triangle tradeoffs (it is notoriously underdefined/specified and could probably be more complete if explicitly correlated to the CAP theorem).

@aschmahmann: during the ACM ICN conference in Macau, we briefly discussed on Zooko's triangle and how we can model naming approaches around it, i.e., the fact that it's difficult (impossible?) to have "one size fits all" solution. Do you remember any more details on this discussion? I think it's going to be useful as we formulate this Open Problem, but I can't recall much from our discussion :)

> What hard constraints should it obey? Are there additional soft constraints that a solution would ideally obey?

A complete solution should repetably demonstrate that it can sustain the load under multiple network conditions for use cases that in the following categories:

- writers (i.e. publishers) to readers (i.e. consumers)
- 1 to many - Blog
- some to many - Code Repository
- writers/readers to writers/readers
- 1 to 1 - Chat
- few to many - Collaborative Documents
- many to many - Social Network, Forums, Chat Rooms

More use cases defined at https://github.com/libp2p/research-pubsub/blob/master/USECASES.md

## Other

### Existing Conversations/Threads

- [IPNS Improvement Design Exploration](https://github.com/ipfs/notes/issues/260)
- [DNS over IPFS](https://github.com/ipfs/notes/issues/250)
- [IPFS database: pubsub, consistency and persistence](https://github.com/ipfs/notes/issues/244)
- [Implement databases over IPFS using Persistent Balanced Trees](https://github.com/ipfs/notes/issues/161)
- [Petnames for multihashes](https://github.com/ipfs/notes/issues/157)
- [pluggable resolvers](https://github.com/ipfs/notes/issues/127)
- [Optimizing IPNS](https://github.com/ipfs/notes/issues/109)
- [Aggregation --> CRDTs discussion](https://github.com/ipfs/notes/issues/40)

### Extra notes