Active-Active EPP HA proposal #1851
Conversation
Signed-off-by: Hang Yin <[email protected]>
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: delavet The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
k8s-ci-robot
added
cncf-cla: yes
|
Hi @delavet. Thanks for your PR. I'm waiting for a github.com member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
k8s-ci-robot
added
the
size/L
Signed-off-by: Hang Yin <[email protected]>
✅ Deploy Preview for gateway-api-inference-extension ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
| # Active-Active HA Deployment Architecture Proposal | ||
| Author(s): @delavet | ||
|
|
||
| ## Proposal Status |
There was a problem hiding this comment.
I'd be happy to see in this proposal some analysis of the tradeoffs -
for example:
- what is the overhead of redis based sync Vs implementing our own gossip and memberlist.
- how much extra time would be spent on synchronization.
- what are the alternatives? e.g., if we can make active passive work with excellent performance and upon a crash of the active make the passive work as expected within a millisecond, does it still worth the described effort?
I went over the proposal - this seems like a lot of complexity and a lot of code.
we need HA in the long term for sure, the question I'm trying to answer if if current proposal is justified.
There was a problem hiding this comment.
We really need the analysis of this part! I will organize the relevant analysis content and data soon and add it to the proposal.
3 participants
This is a proposal aimed at addressing the Active-Active HA issue raised in #1406. The core idea is to implement a distributed state store that guarantees eventual consistency, based on memberlist and CRDT.
We have conducted an initial PoC validation of this solution, and therefore, we are submitting the proposal to the community, looking forward to feedbacks from the reviewers. If there is anything I should know or anything I should do, please feel free to leave a comment.