github.com/fiatjaf/generic-ristretto@v0.0.1/README.md (about) 1 # Ristretto 2 [](http://godoc.org/github.com/dgraph-io/ristretto) 3 [](https://github.com/dgraph-io/ristretto/actions/workflows/ci-ristretto-tests.yml) 4 [](https://github.com/dgraph-io/ristretto/actions/workflows/ci-ristretto-lint.yml) 5 [](https://coveralls.io/github/dgraph-io/ristretto?branch=main) 6 [](https://goreportcard.com/report/github.com/dgraph-io/ristretto) 7 8 Ristretto is a fast, concurrent cache library built with a focus on performance and correctness. 9 10 The motivation to build Ristretto comes from the need for a contention-free 11 cache in [Dgraph][]. 12 13 [Dgraph]: https://github.com/dgraph-io/dgraph 14 15 ## Features 16 17 * **High Hit Ratios** - with our unique admission/eviction policy pairing, Ristretto's performance is best in class. 18 * **Eviction: SampledLFU** - on par with exact LRU and better performance on Search and Database traces. 19 * **Admission: TinyLFU** - extra performance with little memory overhead (12 bits per counter). 20 * **Fast Throughput** - we use a variety of techniques for managing contention and the result is excellent throughput. 21 * **Cost-Based Eviction** - any large new item deemed valuable can evict multiple smaller items (cost could be anything). 22 * **Fully Concurrent** - you can use as many goroutines as you want with little throughput degradation. 23 * **Metrics** - optional performance metrics for throughput, hit ratios, and other stats. 24 * **Simple API** - just figure out your ideal `Config` values and you're off and running. 25 26 ## Status 27 28 Ristretto is production-ready. See [Projects using Ristretto](#projects-using-ristretto). 29 30 ## Table of Contents 31 32 * [Usage](#Usage) 33 * [Example](#Example) 34 * [Config](#Config) 35 * [NumCounters](#Config) 36 * [MaxCost](#Config) 37 * [BufferItems](#Config) 38 * [Metrics](#Config) 39 * [OnEvict](#Config) 40 * [KeyToHash](#Config) 41 * [Cost](#Config) 42 * [Benchmarks](#Benchmarks) 43 * [Hit Ratios](#Hit-Ratios) 44 * [Search](#Search) 45 * [Database](#Database) 46 * [Looping](#Looping) 47 * [CODASYL](#CODASYL) 48 * [Throughput](#Throughput) 49 * [Mixed](#Mixed) 50 * [Read](#Read) 51 * [Write](#Write) 52 * [Projects using Ristretto](#projects-using-ristretto) 53 * [FAQ](#FAQ) 54 55 ## Usage 56 57 ### Example 58 59 ```go 60 func main() { 61 cache, err := ristretto.NewCache(&ristretto.Config{ 62 NumCounters: 1e7, // number of keys to track frequency of (10M). 63 MaxCost: 1 << 30, // maximum cost of cache (1GB). 64 BufferItems: 64, // number of keys per Get buffer. 65 }) 66 if err != nil { 67 panic(err) 68 } 69 70 // set a value with a cost of 1 71 cache.Set("key", "value", 1) 72 73 // wait for value to pass through buffers 74 cache.Wait() 75 76 value, found := cache.Get("key") 77 if !found { 78 panic("missing value") 79 } 80 fmt.Println(value) 81 cache.Del("key") 82 } 83 ``` 84 85 ### Config 86 87 The `Config` struct is passed to `NewCache` when creating Ristretto instances (see the example above). 88 89 **NumCounters** `int64` 90 91 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. 92 93 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. 94 95 **MaxCost** `int64` 96 97 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. 98 99 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. 100 101 MaxCost could be anything as long as it matches how you're using the cost values when calling Set. 102 103 **BufferItems** `int64` 104 105 BufferItems is the size of the Get buffers. The best value we've found for this is 64. 106 107 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. 108 109 **Metrics** `bool` 110 111 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. 112 113 **OnEvict** `func(hashes [2]uint64, value interface{}, cost int64)` 114 115 OnEvict is called for every eviction. 116 117 **KeyToHash** `func(key interface{}) [2]uint64` 118 119 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/dgraph-io/ristretto/blob/master/z/z.go#L19-L41). 120 121 Note that if you want 128bit hashes you should use the full `[2]uint64`, 122 otherwise just fill the `uint64` at the `0` position and it will behave like 123 any 64bit hash. 124 125 **Cost** `func(value interface{}) int64` 126 127 Cost is an optional function you can pass to the Config in order to evaluate 128 item cost at runtime, and only for the Set calls that aren't dropped (this is 129 useful if calculating item cost is particularly expensive and you don't want to 130 waste time on items that will be dropped anyways). 131 132 To signal to Ristretto that you'd like to use this Cost function: 133 134 1. Set the Cost field to a non-nil function. 135 2. When calling Set for new items or item updates, use a `cost` of 0. 136 137 ## Benchmarks 138 139 The benchmarks can be found in https://github.com/dgraph-io/benchmarks/tree/master/cachebench/ristretto. 140 141 ### Hit Ratios 142 143 #### Search 144 145 This trace is described as "disk read accesses initiated by a large commercial 146 search engine in response to various web search requests." 147 148 <p align="center"> 149 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Hit%20Ratios%20-%20Search%20(ARC-S3).svg"> 150 </p> 151 152 #### Database 153 154 This trace is described as "a database server running at a commercial site 155 running an ERP application on top of a commercial database." 156 157 <p align="center"> 158 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Hit%20Ratios%20-%20Database%20(ARC-DS1).svg"> 159 </p> 160 161 #### Looping 162 163 This trace demonstrates a looping access pattern. 164 165 <p align="center"> 166 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Hit%20Ratios%20-%20Glimpse%20(LIRS-GLI).svg"> 167 </p> 168 169 #### CODASYL 170 171 This trace is described as "references to a CODASYL database for a one hour 172 period." 173 174 <p align="center"> 175 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Hit%20Ratios%20-%20CODASYL%20(ARC-OLTP).svg"> 176 </p> 177 178 ### Throughput 179 180 All throughput benchmarks were ran on an Intel Core i7-8700K (3.7GHz) with 16gb 181 of RAM. 182 183 #### Mixed 184 185 <p align="center"> 186 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Throughput%20-%20Mixed.svg"> 187 </p> 188 189 #### Read 190 191 <p align="center"> 192 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Throughput%20-%20Read%20(Zipfian).svg"> 193 </p> 194 195 #### Write 196 197 <p align="center"> 198 <img src="https://raw.githubusercontent.com/dgraph-io/ristretto/master/benchmarks/Throughput%20-%20Write%20(Zipfian).svg"> 199 </p> 200 201 ## Projects Using Ristretto 202 203 Below is a list of known projects that use Ristretto: 204 205 - [Badger](https://github.com/dgraph-io/badger) - Embeddable key-value DB in Go 206 - [Dgraph](https://github.com/dgraph-io/dgraph) - Horizontally scalable and distributed GraphQL database with a graph backend 207 - [Vitess](https://github.com/vitessio/vitess) - Database clustering system for horizontal scaling of MySQL 208 - [SpiceDB](https://github.com/authzed/spicedb) - Horizontally scalable permissions database 209 210 ## FAQ 211 212 ### How are you achieving this performance? What shortcuts are you taking? 213 214 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. 215 216 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. 217 218 ### Is Ristretto distributed? 219 220 No, it's just like any other Go library that you can import into your project and use in a single process.