RegenerativeDistributedCache 1.0.87

RegenerativeDistributedCache

A cache that supports scheduling the regeneration of cache items in the background ahead of their expiry and manages this across a farm of web/service nodes minimising duplicated cache value generation work.

Requires an external (network) cache, a fan out pub/sub message bus, and a distributed locking mechanism (all three of these can be provided by Redis or you might use alternatives for one or more of these such as RabbitMq for messaging). Basic Redis implementations of these are provided in RegenerativeCacheManager.Redis.

• Packages:
  • NuGet.org → RegenerativeDistributedCache
  • NuGet.org → RegenerativeDistributedCache.Redis
• Website: GitHub → mhano/RegenerativeDistributedCache
• Builds: Appveyor → mhano/regenerativedistributedcache
• License: "MIT" License Here, reference: [OpenSource.org → MIT]

RegenerativeDistributedCache.Redis

Basic Redis backed implementations of the interfaces in RegenerativeDistributedCache.Interfaces for an external (network) cache, a fan out pub/sub message bus, and a distributed locking mechanism for use with RegenerativeDistributedCache.RegenerativeCacheManager.

Use a combination of RedisExternalCache, RedisDistributedLockFactory and RedisFanoutBus connected to your existing wiring / configuration of Redis / RedLock or replace components as needed (such as implementing a RabbitMq or MassTransit based IFanOutBus instead of RedisFanOutBus).

CAUTION - The BasicRedisWrapper wraps up the creation of all three (either based on a single Redis connection, or a Redis connection per concern [caching/locking/messaging]) but performs some pretty basic implemention in regards to connecting to Redis, you will want to review the approach carefully prior to using.

RegenerativeCacheManager

Is the cache that supports scheduling the regeneration of cache items in the background ahead of their expiry (and manages this across a farm of web/service nodes).

Each node takes responsibility for regenerating the cache value if it hasn't been sufficiently recently generated, an external global lock is used to ensure only a single node actually calls the generation callback. All nodes are informed when there is a new value in the cache (causing removal of old value from app memory / allowing lazy fetch of updated value from network [Redis] cache store). Nodes which compete for but don't win the global lock and are waiting on an updated value, await a notification message from the lock winning node and then return the value from the network cache (which also populates the local memory cache for faster future retrievals).

Setup:

As a static/shared instance or singleton from your IOC container:

regenerativeCacheManagerSingleton = new RegenerativeCacheManager(
	"mykeyspace", externalCache, distributedLockFactory, fanoutBus)
{
    // Default values:
    CacheExpiryToleranceSeconds = 30, 
    FarmClockToleranceSeconds = 15,
    MinimumForwardSchedulingSeconds = 5,
    TriggerDelaySeconds = 1,
};

Practical Guidance on Time-spans / Timing Settings:

• Refer to full documentation on GitHub.

Use:

var result = regenerativeCacheManagerSingleton.GetOrAdd(
    key: $"{nameof(Item)}:{itemId}", 

    // will not be called if value exists
    generateFunc: () => GetItem(itemId).AsString(),

    // total time regenerating after last GetOrAdd call on current machine
    inactiveRetention: TimeSpan.FromMinutes(30),

    // how frequently to update cache from generateFunc()
    regenerationInterval : TimeSpan.FromMinutes(2)
);

CorrelatedAwaitManager

Allows user to await the receipt of a message based on a key. Allows multiple threads to receive a single copy of a message (often originating remotely).

Typical use is to support multiple local threads receiving a notification from some remote source (such as a fan out message).

CorrelatedAwaitManager receives a copy of all messages delivered to it then delivers to any threads that have setup an awaiter for the specified key value.

Basically a very short lived hyper efficient subscribe mechanism to support coordination within a distributed system, much cheaper than setting up a typical subscriber.

used in RegenerativeCacheManager

Setup:

_remoteBus.Subscribe<TMessage>(m => _singletonCorrelatedAwaitManager.NotifyAwaiters(m));

Use:

TMessage message; // the awaited notification message type.
TSomething result;

if((result = GetSomething(key)) != null) return result;

// CAUTION awaiter must be disposed or cancelled (awaiter.Cancel())
using(var awaiter = _correlatedAwaitManager.CreateAwaiter(key))
{
	// Note you must be certain that the message is being sent after the
	// awaiter has been created (or you could end up waiting forever / timing out).

	// Often this involves a second test as to whether some result became
	// available just now (some time before/after awaiter creation), therefore
	// we should double check it's not available as the notification about it
	// may have been sent before we setup the awaiter.

	if((result = GetSomething(key)) != null) return result;

	// The below or: message = awaiter.Task.Result;
	// or awaiter.Task.Wait(timeOut);
	message = await awaiter.Task.ConfigureAwait(false);
}

// as we have awaited the awaiter we now know the result is available.
return GetSomething(key);

used in RegenerativeCacheManager

License (MIT)

Copyright (c) 2018 Mhano Harkness

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.