内存相关
Q: What’s the meaning of metric “FreeableMemory” on ElastiCache?
# Problem Description======================
I understand that you are confused about these two values, “the memory of Node Type” and “the memory of freeable.” And you want to know which one or other metrics you should concern. If I misunderstood, please feel free to correct me.
# Explanation=================
As you know, when Redis is used as a cache, often, it is handy to let it automatically evict old data as you add a new one. There is a maxmemory setting directive used to configure Redis to use a specified amount of memory of the data set. When the specified amount of memory is reached, it is possible to select among different behaviors, called policies. Redis can return errors for commands that could result in more memory being used, or it can evict some old data in order to return back to the specified limit every time new data is added. To see more information, you can refer to [1].
When you launched a Redis and chose t2.micro in the Node type, the default maxmemory is 5819680 bytes, equivalent to 555 MB. Here is a document that shows the default values for the maxmemory for each node type. The value of maxmemory is the maximum number of bytes available to you for use, data, and other uses, on the node [2].
As for FreeableMemory, it represents the amount of free memory available on the host. This is derived from the RAM, buffers, and cache that the OS reports as freeable[3].
For Redis, there is a metric named “BytesUsedForCache.” This metric gives the exact size of the data stored by Redis or Memcache. This metric is obtained by running info or stats command in the engines[4].
FreeableMemory = RAM +SWAP - ( BytesUsedForCache + Host OS memory usage).
RAM+SWAP is constant. Therefore, if free memory is decreasing and BytesUsedForCache is constant, it means it is getting utilized by OS cache. The thing about OS cache is if free memory is there, OS likes to keep it utilized, but if there is memory pressure in the machine, it can also be freed by the OS to give it to other processes like Redis or Memcache.
To answer your question, “what kind of metrics should I focus on?” My answer is that you should monitor the BytesUsedForCache. For example, you can put a cloudwatch alarm on BytesUsedForCache if it goes above 75% of the maxmemory in the node.
I hope the information above answered your questions.
Should you have any further questions, please feel free to reach back and I will be glad to assist you further.
# References==============================
[1] https://redis.io/topics/lru-cache
[2] Redis Specific Parameters - Redis Node-Type Specific Parameters, https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/ParameterGroups.Redis.html#ParameterGroups.Redis.NodeSpecific
[3] Host-Level Metrics - https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/CacheMetrics.HostLevel.html
[4] Metrics for Redis - https://docs.aws.amazon.com/en_us/AmazonElastiCache/latest/red-ug/CacheMetrics.Redis.html
这次FAQ的概括：
- FreeableMemory这个metric的意思是ElastiCache实例的宿主机的可用内存+可用swap。
- 如果FreeableMemory持续下降，而BytesUsedForCache稳定，那么说明宿主机的OS占用的内存变多了。
- OS有占用资源的倾向，OS是这么设计的。OS会为应用程序释放自己占用的资源，我们不太需要关心OS占用了多少资源，只要确保OS有可用(free)资源就行了。
- ElastiCache实际使用了多少内存是看BytesUsedForCache。
- ElastiCache实例的最大内存大小是和实例的Node Type相关的，参考[2]。
