{"id":21629704,"url":"https://github.com/sfu-dis/ssd-vs-pm","last_synced_at":"2025-04-11T13:51:15.888Z","repository":{"id":38267419,"uuid":"401952977","full_name":"sfu-dis/ssd-vs-pm","owner":"sfu-dis","description":"Cost/performance analysis of index structures on SSD and persistent memory (CIDR 2022)","archived":false,"fork":false,"pushed_at":"2022-06-23T05:03:17.000Z","size":174,"stargazers_count":36,"open_issues_count":0,"forks_count":1,"subscribers_count":5,"default_branch":"master","last_synced_at":"2025-03-25T10:04:34.861Z","etag":null,"topics":["benchmark","cost-performance","cpp","dcpmm","index","optane-dc","persistent-memory","ssd","storage-hierarchy","storage-jungle"],"latest_commit_sha":null,"homepage":"","language":"C++","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/sfu-dis.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2021-09-01T06:17:59.000Z","updated_at":"2024-12-06T16:30:28.000Z","dependencies_parsed_at":"2022-09-07T17:40:11.616Z","dependency_job_id":null,"html_url":"https://github.com/sfu-dis/ssd-vs-pm","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sfu-dis%2Fssd-vs-pm","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sfu-dis%2Fssd-vs-pm/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sfu-dis%2Fssd-vs-pm/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/sfu-dis%2Fssd-vs-pm/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/sfu-dis","download_url":"https://codeload.github.com/sfu-dis/ssd-vs-pm/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248411942,"owners_count":21099031,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["benchmark","cost-performance","cpp","dcpmm","index","optane-dc","persistent-memory","ssd","storage-hierarchy","storage-jungle"],"created_at":"2024-11-25T02:08:31.680Z","updated_at":"2025-04-11T13:51:15.862Z","avatar_url":"https://github.com/sfu-dis.png","language":"C++","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Benchmarking Optane SSD vs Optane DCPMM\n\nHere hosts resources used in our recent CIDR 2022 [paper](http://www.sfu.ca/~kha85/modern-storage-index.pdf) below for comparing SSD and PM indexes, including a benchmark driver adapted from [1], a B-tree implementation based on [2], an on-disk hash table [3], FPTree [4], BzTree [5] and Dash [6].\n\n```\nSSDs Striking Back: The Storage Jungle and Its Implications on Persistent Indexes.\nKaisong Huang, Darien Imai, Tianzheng Wang and Dong Xie.\n12th Annual Conference on Innovative Data Systems Research (CIDR ’22). January 9-12, 2022, Chaminade, USA.\n```\n\n[1] https://github.com/basicthinker/YCSB-C\u003cbr/\u003e\n[2] https://github.com/jeffreyorihuela/b-tree-on-disk\u003cbr/\u003e\n[3] https://github.com/dongx-psu/hashtable\u003cbr/\u003e\n[4] https://github.com/sfu-dis/fptree\u003cbr/\u003e\n[5] https://github.com/sfu-dis/bztree\u003cbr/\u003e\n[6] https://github.com/baotonglu/dash\n\n## Required Flags\n```\n-path \u003cpathname\u003e: Path to the containing folder of btree or hashtable, or to the pool file of PM-based indexes.\n-tree \u003cindexname\u003e: Choose from [btree hashtable bztree dash pibench].\n-load \u003cbool\u003e: truncate files, and load new indexes, default is false.\n-run \u003cbool\u003e: use the preloaded files to run benchmarks, default is false.\n```\n\n## Useful Options\n* `-stride \u003cn\u003e`:The stride when setting CPU affinity, default is 2.\n* `-starting_cpu \u003cn\u003e`: The starting CPU # for affinity, default is 0.\n* `-benchmarkseconds \u003cn\u003e`: Duration of test, default is 20, can also be configured in the spec files.\n* `-buffer_page \u003cn\u003e`: The number of pages for the buffer pool.\n\n\n## Compile and Run:\n**Make sure the git submodule abseil-cpp is initialized.**\n```\ngit submodule update --init\n```\n\n### Run Btree tests\n```\n$ mkdir build; cd build;\n$ cmake .. -DCMAKE_BUILD_TYPE=\u003cDebug|Release\u003e\n$ make\n$ cd Ycsb\n```\nLoad # records. (# = `recordcount` in the `*.spec` file)\n```\n$ ./ycsb -path \u003c/path/to/tree/dir\u003e -tree btree -threads \u003c#threads\u003e -p \u003c/path/to/workload/spec\u003e -load true -buffer_page \u003c#page\u003e\n```\nRun benchmarks.\n```\n$ ./ycsb -path \u003c/path/to/tree/dir\u003e -tree btree -threads \u003c#threads\u003e -p \u003c/path/to/workload/spec\u003e -run true -buffer_page \u003c#page\u003e\n```\nNOTE: spec file examples can be found in /Ycsb/workloads.\n\n### Run hash table tests\n```\n$ mkdir build; cd build;\n$ CXX=clang++ CC=clang cmake .. -DCMAKE_BUILD_TYPE=\u003cDebug|Release\u003e\n$ make\n$ cd Ycsb\n```\nLoad # records. (# = `recordcount` in the `*.spec` file)\n```\n$ ./ycsb -path \u003c/path/to/hashtable/dir\u003e -tree hashtable -threads \u003c#threads\u003e -p \u003c/path/to/workload/spec\u003e -load true -buffer_page \u003c#page\u003e\n```\nRun benchmarks.\n```\n$ ./ycsb -path \u003c/path/to/hashtable/dir\u003e -tree hashtable -threads \u003c#threads\u003e -p \u003c/path/to/workload/spec\u003e -run true -buffer_page \u003c#page\u003e\n```\n\n### Run BzTree tests\nLoad the tree.\n```\n$ ./ycsb -p \u003cspec\u003e -tree bztree -poolsize $POOL_SIZE_IN_BYTES -path $POOLFILE -threads 1 -starting_cpu -load true\n```\nRun benchmarks.\n```\n$./ycsb -benchmarkseconds 60 -p \u003cspec\u003e -tree bztree -path $POOLFILE -threads $THREADS -starting_cpu $STARTING_CPU -stride 2 -run true\n```\n\n### Run Dash tests\nLoad and run in one command.\n```\n$./ycsb -benchmarkseconds 60 -p \u003cspec\u003e -tree dash -poolsize $POOL_SIZE_IN_BYTES -path $POOLFILE -threads $THREADS -starting_cpu $STARTING_CPU -stride 2 -epoch 1024 -load true -run true\n```\n\n### Run tests with PiBench Wrapper\u003cbr/\u003e\nOur benchmark tool is compatible with PiBench wrappers.\u003cbr/\u003e\nLoad the tree.\n```\n$./ycsb -benchmarkseconds 60 -p \u003cspec\u003e -tree pibench -poolsize $POOL_SIZE_IN_BYTES -wrapper $FPWRAPPER -path $POOLFILE -threads $THREADS -starting_cpu $STARTING_CPU -stride 2 -load true\n```\nRun benchmarks.\n```\n$./ycsb -benchmarkseconds 60 -p \u003cspec\u003e -tree pibench -wrapper $TREEWRAPPER -path $POOLFILE -threads $THREADS -starting_cpu $STARTING_CPU -stride 2 -run true\n```\n\n## Troubleshooting\n1. If you ever hit any issue that comes from Dash during compiling, use GCC instead of clang, and add '-march=native -mtune=native' to the compiler flags.\n2. If you see similar errors like this one: `No rule to make target '/usr/lib/x86_64-linux-gnu/libpmemobj.so', needed by 'Ycsb/ycsb'`, please go to `Bztree/CMakeLists.txt` and change line 19 and line 69 to your own PMDK lib path.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsfu-dis%2Fssd-vs-pm","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsfu-dis%2Fssd-vs-pm","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsfu-dis%2Fssd-vs-pm/lists"}