Skip to content

Commit b0ce7f5

Browse files
authored
Merge pull request #282051 from TheovanKraay/java-doc-updates
Java Ecosystem doc updates
2 parents 9dadcac + 39668ed commit b0ce7f5

File tree

2 files changed

+199
-6
lines changed

2 files changed

+199
-6
lines changed

articles/cosmos-db/nosql/conceptual-resilient-sdk-applications.md

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -110,6 +110,12 @@ From the client perspective, any retries will affect the end to end latency of a
110110

111111
Azure Cosmos DB SDKs provide detailed information in their logs and diagnostics that can help identify which retries are taking place. For more information, see [how to collect .NET SDK diagnostics](troubleshoot-dotnet-sdk-slow-request.md#capture-diagnostics) and [how to collect Java SDK diagnostics](troubleshoot-java-sdk-v4.md#capture-the-diagnostics).
112112

113+
## How can I mitigate retry latency?
114+
115+
Depending on the [circumstances](troubleshoot-sdk-availability.md), in most cases the SDK will route requests to either the local region, the write region (in a single-region write scenario) or the first region in the [preferred regions](tutorial-global-distribution.md#preferred-locations) list. This prioritization minimizes latency in healthy scenarios by primarily connecting to the nearest or most optimal data center.
116+
117+
However, this prioritization also means that requests which are going to result in failure will always be tried in one specific region first for a given error scenario. If failover to another region is preferred in that scenario, this is typically handled at the infrastructure (traffic manager) layer rather than at the SDK level. Proper setup and configuration of your infrastructure can ensure that traffic is rerouted efficiently during regional outages, thereby mitigating the latency that can come with cross-region retries in an outage scenario. For more detailed information on setting up infrastructure-level failover, you can refer to [Azure Traffic Manager documentation](../../traffic-manager/index.yml). Some SDKs support implementing similar failover strategies directly at the SDK level. For example, see [high availability for Java SDK](performance-tips-java-sdk-v4.md#high-availability).
118+
113119
## What about regional outages?
114120

115121
The Azure Cosmos DB SDKs cover regional availability and can perform retries on another account regions. Refer to the [multiregional environments retry scenarios and configurations](troubleshoot-sdk-availability.md) article to understand which scenarios involve other regions.

0 commit comments

Comments
 (0)