[
https://issues.apache.org/jira/browse/IOTDB-648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17105167#comment-17105167
]
Xiangdong Huang commented on IOTDB-648:
---------------------------------------
Hi,
> How does it uses IoTDB cluster thrift? Could you provide a pointer to docs?
IoTDB just uses Thrfit to communicate with each other nodes. There is no too
many difference with current IOTDB implementation. But for performance
consideration, we use the async mode. The async mode is provided by Thrift, so
we have to use and obey its API.
You can get examples from the
org.apache.iotdb.cluster.server.handlers.caller.*.java on "cluster_new" branch.
> [gRPC/Thrift] Proxy Framework
> -----------------------------
>
> Key: IOTDB-648
> URL: https://issues.apache.org/jira/browse/IOTDB-648
> Project: Apache IoTDB
> Issue Type: Wish
> Components: Client/CLI, Client/Java, Client/Others, Client/Python
> Reporter: Julian Feinauer
> Priority: Major
>
> This came up during reasoning about a Ditto "History Store" based on IoTDB,
> see https://github.com/eclipse/ditto/issues/545.
> We would rely on Ditto for managing permissions (as it already does for the
> "live" data) so we would need a way to combine that. My idea was to create a
> gRPC/Thrift Proxy / Intersector which registers as Server and interacts with
> the "real" clinet and builds a connection to the "real" server, a man in the
> middle. He then gets the Query and can check it for Ditto permissions (but we
> could keep it generic for ohter use cases too) and only pass it, if its okay.
> What do others think about that?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
