AWS AmazonElastiCache: Troubleshooting updates for public endpoints and TLS 1.3
Summary
Updates TLS, VPC, and security group troubleshooting guidance to distinguish public endpoint caches (TLS 1.3 required, no VPC/security groups) from VPC caches, and adds latency best practices about primary node AZ changes and distributing clients across AZs.
Security assessment
Clarifies encryption-in-transit and network access expectations for public endpoint caches (TLS 1.3, no security groups), which is security-relevant guidance but not tied to a specific vulnerability.
Evidence
For serverless caches with a public endpoint, your client must support TLS 1.3.
Diff
diff --git a/AmazonElastiCache/latest/dg/wwe-troubleshooting.md b/AmazonElastiCache/latest/dg/wwe-troubleshooting.md index d773157..a7baf10 100644 --- a//AmazonElastiCache/latest/dg/wwe-troubleshooting.md +++ b//AmazonElastiCache/latest/dg/wwe-troubleshooting.md @@ -36 +36 @@ If you are unable to connect to your ElastiCache cache, consider one of the foll - 1. **Using TLS:** If you are experiencing a hung connection when trying to connect to your ElastiCache endpoint, you may not be using TLS in your client. If you are using ElastiCache Serverless, encryption in transit is always enabled. Make sure that your client is using TLS to connect to the cache. [Learn more about connecting to a TLS enabled cache](./connect-tls.html). + 1. **Using TLS:** If you are experiencing a hung connection when trying to connect to your ElastiCache endpoint, you may not be using TLS in your client. If you are using ElastiCache Serverless, encryption in transit is always enabled. Make sure that your client is using TLS to connect to the cache. For serverless caches with a public endpoint, your client must support TLS 1.3. [Learn more about connecting to a TLS enabled cache](./connect-tls.html). @@ -38 +38 @@ If you are unable to connect to your ElastiCache cache, consider one of the foll - 2. **VPC:** ElastiCache caches are accessible only from within a VPC. Ensure that the EC2 instance from which you are accessing the cache and the ElastiCache cache are created in the same VPC. Alternatively, you must enable [VPC peering](https://docs.aws.amazon.com/vpc/latest/peering/what-is-vpc-peering.html) between the VPC where your EC2 instance resides and the VPC where you are creating your cache. + 2. **VPC:** ElastiCache node-based clusters and Serverless caches with a VPC endpoint are accessible from within a VPC. For these caches, ensure that the EC2 instance from which you are accessing the cache and the ElastiCache cache are created in the same VPC. Alternatively, you must enable [VPC peering](https://docs.aws.amazon.com/vpc/latest/peering/what-is-vpc-peering.html) between the VPC where your EC2 instance resides and the VPC where you are creating your cache. Serverless caches with a public endpoint are accessible over the internet and do not require this VPC configuration. @@ -40 +40 @@ If you are unable to connect to your ElastiCache cache, consider one of the foll - 3. **Security groups:** ElastiCache uses security groups to control access to your cache. Consider the following: + 3. **Security groups:** For caches with a VPC endpoint, ElastiCache uses security groups to control access to your cache. Caches with a public endpoint do not use security groups. For VPC caches, consider the following: @@ -86,0 +87,4 @@ Consider the following best practices and strategies to reduce latency: + * **Account for changes to the primary node's AZ:** After events such as patching or a failover, ElastiCache Serverless can promote a node in a different AZ to primary. Writes, and reads that are not served from a replica, always go to the primary. If your client runs in a single AZ, these requests were previously local to the primary, and after the change all of them cross to the remote AZ, which raises latency. You see this as a step increase in `SuccessfulReadRequestLatency` and `SuccessfulWriteRequestLatency` that persists until the primary moves again. This latency increase is expected behavior for ElastiCache Serverless, and you can expect it to stay within the normal range. + + * **Distribute your client across multiple AZs:** Deploy your client in two or more of your cache's AZs. This routes a portion of your writes and reads served from the primary locally, which reduces average latency. A client spread evenly across two AZs routes about half of these requests locally. +