Azure Cache for Redis management FAQs

Redis server doesn't natively support TLS, but Azure Cache for Redis does. If you're connecting to Azure Cache for Redis and your client supports TLS, like StackExchange.Redis, then use TLS.

Redis tools such as redis-cli don't work with the TLS port, but you can use a utility such as stunnel to securely connect the tools to the TLS port by following the directions in the Announcing ASP.NET Session State Provider for Redis Preview Release blog post.

For instructions on downloading the Redis tools, see the How can I run Redis commands? section.

• StackExchange.Redis best practices
• Configuration and concepts
• Performance testing

StackExchange.Redis best practices

• Set AbortConnect to false, then let the ConnectionMultiplexer reconnect automatically.
• Use a single, long-lived ConnectionMultiplexer instance rather than creating a new connection for each request. For an example of how to manage a connection, see the RedisConnection class in Connect to the cache by using RedisConnection.
• Redis works best with smaller values, so consider chopping up bigger data into multiple keys. For more information, see Best practices development.
• Configure your ThreadPool settings to avoid timeouts.
• Use at least the default connectTimeout of 5 seconds. This interval gives StackExchange.Redis sufficient time to re-establish the connection if there's a network blip.
• Be aware of the performance costs associated with different operations you're running. For instance, the KEYS command is an O(n) operation and should be avoided. The redis.io site has details around the time complexity for each operation that it supports. Select each command to see the complexity for each operation.

Configuration and concepts

• Use Standard or Premium Tier for Production systems. The Basic Tier is a single node system with no data replication and no SLA. Also, use at least a C1 cache. C0 caches are typically used for simple dev/test scenarios.
• Remember that Redis is an In-Memory data store. For more information, see Troubleshoot data loss in Azure Cache for Redis so that you're aware of scenarios where data loss can occur.
• Develop your system such that it can handle connection blips caused by patching and failover.

Performance testing

• Start by using redis-benchmark.exe to get a feel for possible throughput before writing your own perf tests. Because redis-benchmark doesn't support TLS, you must enable the Non-TLS port through the Azure portal before you run the test. For examples, see How can I benchmark and test the performance of my cache?
• The client VM used for testing should be in the same region as your Azure Cache for Redis instance.
• We recommend using Dv2 VM Series for your client as they have better hardware and should give the best results.
• Make sure the client VM you choose has at least as much computing and bandwidth capability as the cache you're testing.
• Enable VRSS on the client machine if you are on Windows. See here for details.
• Premium tier Redis instances have better network latency and throughput because they're running on better hardware for both CPU and Network.

• Avoid using certain Redis commands that take a long time to complete, unless you fully understand the result of these commands. For example, don't run the KEYS command in production. Depending on the number of keys, it could take a long time to return. Redis is a single-threaded server and it processes commands one at a time. If you have other commands issued after KEYS, they're not be processed until Redis processes the KEYS command. The redis.io site has details around the time complexity for each operation that it supports. Select each command to see the complexity for each operation.
• Key sizes - should I use small key/values or large key/values? It depends on the scenario. If your scenario requires larger keys, you can adjust the ConnectionTimeout, then retry values and adjust your retry logic. From a Redis server perspective, smaller values give better performance.
• These considerations don't mean that you can't store larger values in Redis; you must be aware of the following considerations. Latencies will be higher. If you have one set of data that is larger and one that is smaller, you can use multiple ConnectionMultiplexer instances. Configure each with a different set of timeout and retry values, as described in the previous What do the StackExchange.Redis configuration options do section.

• Enable cache diagnostics so you can monitor the health of your cache. You can view the metrics in the Azure portal and you can also download and review them using the tools of your choice.
• You can use redis-benchmark.exe to load test your Redis server.
• Ensure that the load testing client and the Azure Cache for Redis are in the same region.
• Use redis-cli.exe and monitor the cache using the INFO command.
• If your load is causing high memory fragmentation, you should scale up to a larger cache size.
• For instructions on downloading the Redis tools, see the How can I run Redis commands? section.

Here are some examples of using redis-benchmark.exe. Run these commands from a VM in the same region as your cache for accurate results.cache-development-faq.yml

• Test Pipelined SET requests using a 1k payload

  redis-benchmark.exe -h **yourcache**.redis.cache.chinacloudapi.cn -a **yourAccesskey** -t SET -n 1000000 -d 1024 -P 50

• Test Pipelined GET requests using a 1k payload.

redis-benchmark.exe -h **yourcache**.redis.cache.chinacloudapi.cn -a **yourAccesskey** -t GET -n 1000000 -d 1024 -P 50

The CLR ThreadPool has two types of threads - "Worker" and "I/O Completion Port" (IOCP) threads.

• Worker threads are used for things like processing the Task.Run(…), or ThreadPool.QueueUserWorkItem(…) methods. These threads are also used by various components in the CLR when work needs to happen on a background thread.
• IOCP threads are used when asynchronous IO happens, such as when reading from the network.

The thread pool provides new worker threads or I/O completion threads on demand (without any throttling) until it reaches the "Minimum" setting for each type of thread. By default, the minimum number of threads is set to the number of processors on a system.

Once the number of existing (busy) threads hits the "minimum" number of threads, the ThreadPool will throttle the rate at which it injects new threads to one thread per 500 milliseconds. Typically, if your system gets a burst of work needing an IOCP thread, it will process that work quickly. However, if the burst is more than the configured "Minimum" setting, there's some delay in processing some of the work as the ThreadPool waits for one of two possibilities:

• An existing thread becomes free to process the work.
• No existing thread becomes free for 500 ms and a new thread is created.

Basically, when the number of Busy threads is greater than Min threads, you're likely paying a 500-ms delay before network traffic is processed by the application. Also, when an existing thread stays idle for longer than 15 seconds, it's cleaned up and this cycle of growth and shrinkage can repeat.

If we look at an example error message from StackExchange.Redis (build 1.0.450 or later), we see that it now prints ThreadPool statistics. See IOCP and WORKER details below.

System.TimeoutException: Timeout performing GET MyKey, inst: 2, mgr: Inactive,
queue: 6, qu: 0, qs: 6, qc: 0, wr: 0, wq: 0, in: 0, ar: 0,
IOCP: (Busy=6,Free=994,Min=4,Max=1000),
WORKER: (Busy=3,Free=997,Min=4,Max=1000)

As shown In the example, you see that for IOCP thread there are six busy threads and the system is configured to allow four minimum threads. In this case, the client would have likely seen two 500-ms delays, because 6 > 4.

Recommendation

Given this information, we strongly recommend that customers set the minimum configuration value for IOCP and WORKER threads to something larger than the default value. We can't give one-size-fits-all guidance on what this value should be because the right value for one application will likely be too high or low for another application. This setting can also affect the performance of other parts of complicated applications, so each customer needs to fine-tune this setting to their specific needs. A good starting place is 200 or 300, then test and tweak as needed.

How to configure this setting:

• We recommend changing this setting programmatically by using the ThreadPool.SetMinThreads (...) method in global.asax.cs. For example:

  private readonly int minThreads = 200;
  void Application_Start(object sender, EventArgs e)
  {
      // Code that runs on application startup
      AreaRegistration.RegisterAllAreas();
      RouteConfig.RegisterRoutes(RouteTable.Routes);
      BundleConfig.RegisterBundles(BundleTable.Bundles);
      ThreadPool.SetMinThreads(minThreads, minThreads);
  }
  

  Note

  The value specified by this method is a global setting, affecting the whole AppDomain. For example, if you have a 4-core machine and want to set minWorkerThreads and minIoThreads to 50 per CPU during run-time, use ThreadPool.SetMinThreads(200, 200).

• It is also possible to specify the minimum threads setting by using the minIoThreads or minWorkerThreads configuration setting under the <processModel> configuration element in Machine.config. Machine.config is typically located at %SystemRoot%\Microsoft.NET\Framework\[versionNumber]\CONFIG\. Setting the number of minimum threads in this way isn't recommended because it's a System-wide setting.

  Note

  The value specified in this configuration element is a per-core setting. For example, if you have a 4-core machine and want your minIoThreads setting to be 200 at runtime, you would use <processModel minIoThreads="50"/>.

Enabling server GC can optimize the client and provide better performance and throughput when using StackExchange.Redis. For more information on server GC and how to enable it, see the following articles:

• To enable server GC
• Fundamentals of Garbage Collection
• Garbage Collection and Performance

Each pricing tier has different limits for client connections, memory, and bandwidth. While each size of cache allows up to some number of connections, each connection to Redis has overhead associated with it. An example of such overhead would be CPU and memory usage because of TLS/SSL encryption. The maximum connection limit for a given cache size assumes a lightly loaded cache. If load from connection overhead plus load from client operations exceeds capacity for the system, the cache can experience capacity issues even if you haven't exceeded the connection limit for the current cache size.

For more information about the different connections limits for each tier, see Azure Cache for Redis pricing. For more information about connections and other default configurations, see Default Redis server configuration.

Next steps

Learn about other Azure Cache for Redis FAQs.