github.com/etecs-ru/ristretto@v0.9.1/README.md (about) 1 # Ristretto 2 [](https://pkg.go.dev/github.com/etecs-ru/ristretto) 3 [](https://goreportcard.com/report/github.com/etecs-ru/ristretto) 4 [](https://github.com/etecs-ru/ristretto/releases/latest) 5 [](https://codecov.io/gh/etecs-ru/ristretto) 6  7 8 9 10 ## This is the fork of the [github.com/dgraph-io/ristretto](https://github.com/dgraph-io/ristretto). 11 12 This fork applies the selected PRs from the original repo. Also runs GitHub Actions on each PR. See the CHANGELOG.md. 13 14 ---- 15 16 17 Ristretto is a fast, concurrent cache library built with a focus on performance and correctness. 18 19 The motivation to build Ristretto comes from the need for a contention-free 20 cache in [Dgraph][]. 21 22 **Use [Discuss Issues](https://discuss.dgraph.io/tags/c/issues/35/ristretto/40) for reporting issues about this repository.** 23 24 [Dgraph]: https://github.com/dgraph-io/dgraph 25 26 ## Features 27 28 * **High Hit Ratios** - with our unique admission/eviction policy pairing, Ristretto's performance is best in class. 29 * **Eviction: SampledLFU** - on par with exact LRU and better performance on Search and Database traces. 30 * **Admission: TinyLFU** - extra performance with little memory overhead (12 bits per counter). 31 * **Fast Throughput** - we use a variety of techniques for managing contention and the result is excellent throughput. 32 * **Cost-Based Eviction** - any large new item deemed valuable can evict multiple smaller items (cost could be anything). 33 * **Fully Concurrent** - you can use as many goroutines as you want with little throughput degradation. 34 * **Metrics** - optional performance metrics for throughput, hit ratios, and other stats. 35 * **Simple API** - just figure out your ideal `Config` values and you're off and running. 36 37 ## Status 38 39 Ristretto is production-ready. See [Projects using Ristretto](#projects-using-ristretto). 40 41 ## Table of Contents 42 43 * [Usage](#Usage) 44 * [Example](#Example) 45 * [Config](#Config) 46 * [NumCounters](#Config) 47 * [MaxCost](#Config) 48 * [BufferItems](#Config) 49 * [Metrics](#Config) 50 * [OnEvict](#Config) 51 * [KeyToHash](#Config) 52 * [Cost](#Config) 53 * [Testing](#Testing) 54 * [Benchmarks](#Benchmarks) 55 * [Hit Ratios](#Hit-Ratios) 56 * [Search](#Search) 57 * [Database](#Database) 58 * [Looping](#Looping) 59 * [CODASYL](#CODASYL) 60 * [Throughput](#Throughput) 61 * [Mixed](#Mixed) 62 * [Read](#Read) 63 * [Write](#Write) 64 * [Projects using Ristretto](#projects-using-ristretto) 65 * [FAQ](#FAQ) 66 67 68 ## Usage 69 70 ### Example 71 72 ```go 73 func main() { 74 cache, err := ristretto.NewCache(&ristretto.Config{ 75 NumCounters: 1e7, // number of keys to track frequency of (10M). 76 MaxCost: 1 << 30, // maximum cost of cache (1GB). 77 BufferItems: 64, // number of keys per Get buffer. 78 }) 79 if err != nil { 80 panic(err) 81 } 82 83 // set a value with a cost of 1 84 cache.Set("key", "value", 1) 85 86 // wait for value to pass through buffers 87 cache.Wait() 88 89 value, found := cache.Get("key") 90 if !found { 91 panic("missing value") 92 } 93 fmt.Println(value) 94 cache.Del("key") 95 } 96 ``` 97 98 ### Config 99 100 The `Config` struct is passed to `NewCache` when creating Ristretto instances (see the example above). 101 102 **NumCounters** `int64` 103 104 NumCounters is the number of 4-bit access counters to keep for admission and eviction. We've seen good performance in setting this to 10x the number of items you expect to keep in the cache when full. 105 106 For example, if you expect each item to have a cost of 1 and MaxCost is 100, set NumCounters to 1,000. Or, if you use variable cost values but expect the cache to hold around 10,000 items when full, set NumCounters to 100,000. The important thing is the *number of unique items* in the full cache, not necessarily the MaxCost value. 107 108 **MaxCost** `int64` 109 110 MaxCost is how eviction decisions are made. For example, if MaxCost is 100 and a new item with a cost of 1 increases total cache cost to 101, 1 item will be evicted. 111 112 MaxCost can also be used to denote the max size in bytes. For example, if MaxCost is 1,000,000 (1MB) and the cache is full with 1,000 1KB items, a new item (that's accepted) would cause 5 1KB items to be evicted. 113 114 MaxCost could be anything as long as it matches how you're using the cost values when calling Set. 115 116 **BufferItems** `int64` 117 118 BufferItems is the size of the Get buffers. The best value we've found for this is 64. 119 120 If for some reason you see Get performance decreasing with lots of contention (you shouldn't), try increasing this value in increments of 64. This is a fine-tuning mechanism and you probably won't have to touch this. 121 122 **Metrics** `bool` 123 124 Metrics is true when you want real-time logging of a variety of stats. The reason this is a Config flag is because there's a 10% throughput performance overhead. 125 126 **OnEvict** `func(hashes [2]uint64, value interface{}, cost int64)` 127 128 OnEvict is called for every eviction. 129 130 **KeyToHash** `func(key interface{}) [2]uint64` 131 132 KeyToHash is the hashing algorithm used for every key. If this is nil, Ristretto has a variety of [defaults depending on the underlying interface type](https://github.com/etecs-ru/ristretto/blob/master/z/z.go#L19-L41). 133 134 Note that if you want 128bit hashes you should use the full `[2]uint64`, 135 otherwise just fill the `uint64` at the `0` position and it will behave like 136 any 64bit hash. 137 138 **Cost** `func(value interface{}) int64` 139 140 Cost is an optional function you can pass to the Config in order to evaluate 141 item cost at runtime, and only for the Set calls that aren't dropped (this is 142 useful if calculating item cost is particularly expensive and you don't want to 143 waste time on items that will be dropped anyways). 144 145 To signal to Ristretto that you'd like to use this Cost function: 146 147 1. Set the Cost field to a non-nil function. 148 2. When calling Set for new items or item updates, use a `cost` of 0. 149 150 ### Testing 151 152 If you wish to mock out the caching functionality for use in your tests, you should use the `CacheInterface` in your code. 153 This enables you to generate mocks using something like [mockery](https://github.com/vektra/mockery) or provide your own implementation. 154 155 ## Benchmarks 156 157 The benchmarks can be found in https://github.com/dgraph-io/benchmarks/tree/master/cachebench/ristretto. 158 159 ### Hit Ratios 160 161 #### Search 162 163 This trace is described as "disk read accesses initiated by a large commercial 164 search engine in response to various web search requests." 165 166 <p align="center"> 167 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Hit%20Ratios%20-%20Search%20(ARC-S3).svg"> 168 </p> 169 170 #### Database 171 172 This trace is described as "a database server running at a commercial site 173 running an ERP application on top of a commercial database." 174 175 <p align="center"> 176 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Hit%20Ratios%20-%20Database%20(ARC-DS1).svg"> 177 </p> 178 179 #### Looping 180 181 This trace demonstrates a looping access pattern. 182 183 <p align="center"> 184 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Hit%20Ratios%20-%20Glimpse%20(LIRS-GLI).svg"> 185 </p> 186 187 #### CODASYL 188 189 This trace is described as "references to a CODASYL database for a one hour 190 period." 191 192 <p align="center"> 193 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Hit%20Ratios%20-%20CODASYL%20(ARC-OLTP).svg"> 194 </p> 195 196 ### Throughput 197 198 All throughput benchmarks were ran on an Intel Core i7-8700K (3.7GHz) with 16gb 199 of RAM. 200 201 #### Mixed 202 203 <p align="center"> 204 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Throughput%20-%20Mixed.svg"> 205 </p> 206 207 #### Read 208 209 <p align="center"> 210 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Throughput%20-%20Read%20(Zipfian).svg"> 211 </p> 212 213 #### Write 214 215 <p align="center"> 216 <img src="https://raw.githubusercontent.com/etecs-ru/ristretto/master/benchmarks/Throughput%20-%20Write%20(Zipfian).svg"> 217 </p> 218 219 ## Projects Using Ristretto 220 221 Below is a list of known projects that use Ristretto: 222 223 - [Badger](https://github.com/dgraph-io/badger) - Embeddable key-value DB in Go 224 - [Dgraph](https://github.com/dgraph-io/dgraph) - Horizontally scalable and distributed GraphQL database with a graph backend 225 - [Vitess](https://github.com/vitessio/vitess) - Database clustering system for horizontal scaling of MySQL 226 - [SpiceDB](https://github.com/authzed/spicedb) - Horizontally scalable permissions database 227 228 ## FAQ 229 230 ### How are you achieving this performance? What shortcuts are you taking? 231 232 We go into detail in the [Ristretto blog post](https://blog.dgraph.io/post/introducing-ristretto-high-perf-go-cache/), but in short: our throughput performance can be attributed to a mix of batching and eventual consistency. Our hit ratio performance is mostly due to an excellent [admission policy](https://arxiv.org/abs/1512.00727) and SampledLFU eviction policy. 233 234 As for "shortcuts," the only thing Ristretto does that could be construed as one is dropping some Set calls. That means a Set call for a new item (updates are guaranteed) isn't guaranteed to make it into the cache. The new item could be dropped at two points: when passing through the Set buffer or when passing through the admission policy. However, this doesn't affect hit ratios much at all as we expect the most popular items to be Set multiple times and eventually make it in the cache. 235 236 ### Is Ristretto distributed? 237 238 No, it's just like any other Go library that you can import into your project and use in a single process.