Project

General

Profile

Actions

Feature #4762

open

Add a new UnsolicitedDataPolicy that admits unsolicited data by prefix

Added by Ernest McCracken over 5 years ago. Updated over 2 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
Forwarding
Target version:
-
Start date:
Due date:
% Done:

0%

Estimated time:

Description

We would like to add an unsolicited data policy that allows applications to specify which prefixes they would like to cache unsolicited data for.

The proposed design is as follows:
Create a new UnsolicityDataPolcy called AdmitPrefixesUnsolicitedDataPolicy that defines two additional methods register(name) and unregister(name).
Prefixes are stored in a table which is looked up in the decide method.

Create a new manager class to allow applications to use the Interest based API to add/remove prefixes as needed.


Files

f4b9a96.diff.zip (4.33 KB) f4b9a96.diff.zip Change 5517,17 Junxiao Shi, 01/11/2022 02:44 PM
Actions #1

Updated by Davide Pesavento over 5 years ago

  • Subject changed from Add a new UnsolicitedDataPolicy that admits unsolicited data by prefix. to Add a new UnsolicitedDataPolicy that admits unsolicited data by prefix
  • Start date deleted (10/23/2018)
Actions #2

Updated by Ernest McCracken over 5 years ago

I've mostly finished the changes needed for the UnsolicitedDataPolicy implementation. I do have questions about management.

The choice of UnsolicitedDataPolicy is made in the config file. Is there any benefit to changing the policy while NFD is running?

If not then does it make sense for the content store management module to handle requests to add and remove prefixes from the AdmitPrefixesUnsolicitedDataPolicy (if that is the current policy) instead of creating a new management module just for UnsolicitedDataPolicy?

Actions #3

Updated by Ernest McCracken over 5 years ago

  • Status changed from New to In Progress
Actions #4

Updated by Ernest McCracken over 5 years ago

Some Issues to think about:

  • The number or prefixes that might be unsolicited. If the order reaches into 1000's then a name tree instead of unordered_set would be more appropriate. This may be something that simply needs to be monitored once the feature is being used.

  • Use of regular expressions. Beneficial from a machine learning perspective as well as reducing the size of unsolicited prefixes.

Actions #5

Updated by Davide Pesavento almost 4 years ago

What's the status of this issue?

Actions #6

Updated by Lan Wang almost 4 years ago

Davide Pesavento wrote:

What's the status of this issue?

Ernest implemented this feature but the code has not been reviewed or merged. If there's interest in the code, I can ask Alex to take on this task after he's done with the ASF changes.

Actions #7

Updated by Davide Pesavento about 3 years ago

  • Category changed from Management to Forwarding
  • Status changed from In Progress to New
  • Assignee deleted (Ernest McCracken)
Actions #9

Updated by Davide Pesavento over 2 years ago

Junxiao Shi wrote in #note-8:

https://gerrit.named-data.net/c/NFD/+/5517

What's that? I get a 404

Actions #10

Updated by Junxiao Shi over 2 years ago

Davide Pesavento wrote in #note-9:

Junxiao Shi wrote in #note-8:

https://gerrit.named-data.net/c/NFD/+/5517

What's that? I get a 404

It's Jeremy's commit.
Maybe some accounts cannot see it.
Therefore I've uploaded the patch file.

Actions

Also available in: Atom PDF