[
https://issues.apache.org/jira/browse/IGNITE-9676?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16643605#comment-16643605
]
Vladimir Ozerov commented on IGNITE-9676:
-----------------------------------------
[[email protected]], [~antkr], [~agerus],
Igniters,
I think we should move this ticket to AI 2.8 release. It seems that at this
point we do not have the whole picture on how this stuff is expected to be used
by users, and certain design decisions looks suspiciously. The ultimate goal is
to have implementation which is on par with our competitors (GemFire,
Hazelcast, etc.)
Namely:
1) We need to better handle expirations - every read should update entry's
access time to prevent too early expiry. Also we'd better to have automatic
expiration on the server side with help of Expiry policy
2) Currently we use either Ignite client node or REST. I think we'd better to
user Ignite client node or thin client, as it is more performant and is under
active development.
3) Another question is how to configure all the stuff properly. E.g. currently
we always create REPLICATED cache for Ignite node case, which seems to be wrong
(unscalable).
4) Other vendors typically save deltas, not the whole session. We also did this
for servlet web sessions. Seems to be a good optimization.
5) Other vendors do some magic with special Spring security attributes. We need
to understand what is this, and implement on our side if needed.
> Ignite as storage in Spring Session
> -----------------------------------
>
> Key: IGNITE-9676
> URL: https://issues.apache.org/jira/browse/IGNITE-9676
> Project: Ignite
> Issue Type: New Feature
> Components: spring
> Reporter: Anton Kurbanov
> Assignee: Anton Kurbanov
> Priority: Minor
> Fix For: 2.8
>
>
> Implement repository backed with Ignite for sessions clustering with Spring
> Session.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
