Micron Technology (kamfani ne na musamman kan samar da DRAM da kuma ƙwaƙwalwar ajiya) fito da gabatarwar wani sabon inji ake kira "HSE" (Injin Injin Ma'aurata iri-iri), wanda aka haɓaka tare da ƙayyadaddun abubuwan amfani a hankali a kan NAND flash based SSD drives (X100, TLC, QLC 3D NAND) ko ƙwaƙwalwar ajiya kawai (NVDIMM).
Injin ana kera shi ne ta hanyar dakin karatu saka cikin wasu aikace-aikace da yana tallafawa sarrafa bayanai a cikin tsari mai ƙima-ƙimar. An rubuta lambar HSE a cikin C kuma an rarraba shi a ƙarƙashin lasisin Apache 2.0.
Daga cikin aikace-aikacen injiniya, aikace-aikace don adana bayanan ƙananan bayanai an ambata a cikin NoSQL DBMS, ɗakunan ajiya na software (SDS, ajiyar ajiyar software) kamar Ceph da Scality RING, dandamali don sarrafa bayanai mai yawa (Big Data), tsarin sarrafa kwamfuta mai ƙarfi (HPC), Intanet na abubuwa (IoT) na'urorin da mafita ga tsarin koyon na'ura.
- HSE an inganta shi ba kawai don iyakar aikin ba, amma kuma don tabbatar da karko na nau'ikan nau'ikan tafiyarwar SSD. An sami babban gudunta hanyar samfurin ajiya samfurin: an adana bayanan da suka fi dacewa, yana rage adadin hanyoyin samun faifai.
A matsayin misali na haɗa sabon injin a cikin ayyukan ɓangare na uku, an shirya sigar da ke kan takardu na MongoDB DBMS, wanda aka fassara don amfani da HSE.
Ta hanyar fasaha, HSE ya dogara ne akan ƙarin ƙirar ƙirar kernel mpool, wanda ke aiwatar da keɓaɓɓen keɓaɓɓen don adana abubuwa don tafiyarwa mai ƙarfi, la'akari da ƙwarewarsu da halayensu, yana ba da damar halaye daban-daban na sauri da karko. Mpool shima wani ƙirar fasaha ce ta Micron yana buɗewa lokaci ɗaya tare da HSE, amma ya zama fitacce a zaman babban aikin more rayuwa. Mpool yana ɗaukar amfani da ƙwaƙwalwar ajiya mai ɗorewa da ajiyar yanki, amma a halin yanzu SSDs na gargajiya kawai ake tallafawa.
Gwajin aiki tare da kunshin YCSB (Yahoo Cloud Serving Benchmark) ya nuna haɓaka mai yawa a yayin amfani da ajiyar 2TB tare da sarrafa toshe bayanan 1KB. Ana lura da haɓakar haɓaka na musamman a cikin gwajin tare da rarraba rarraba ayyukan karatu da rubutu.
Alal misali, MongoDB tare da injin HSE ya zama kusan sau 8 da sauri fiye da sigar tare da daidaitaccen injin WiredTiger, kuma injin RocksDB DBMS ya fi ƙarfin HSE fiye da sau 6. Hakanan ana iya samun kyawawan alamu a cikin gwaje-gwajen, wanda ke nuna kashi 95% na ayyukan karantawa da 5% na canje-canje ko ƙari.
Wani gwajin da aka yi ya ƙunshi ayyukan karantawa kawai, yana nuna fa'idar kusan 40%. Inara yawan yiwuwar SSDs yayin ayyukan rubutawa idan aka kwatanta da tushen tushen RocksDB an kiyasta shi sau 7.
Babban fasali na HSE:
- Taimako don daidaitattun masu ci gaba aiwatar da bayanai a cikin maɓalli / ƙimar tsari;
- Cikakken tallafi na ma'amala tare da ikon keɓe sassan ajiya ta ƙirƙirar hotunan hoto (ana iya amfani da hotunan hoto don adana tarin daban a cikin shago).
- Ikon amfani da siginan sigina don ratsa bayanai a cikin wakilcin hoto.
- Misalin bayanan da aka inganta don nau'ikan nau'ikan kayan haɗi a cikin ma'ajiyar ajiya guda.
- Hanyoyi masu sassauci don sarrafa amincin ajiya.
- Tsarin keɓaɓɓiyar mahimmin tsarin bayanai (rarrabawa ta hanyar nau'ikan ƙwaƙwalwar ajiya da ke cikin maɓallin ajiya).
- Laburare tare da C API wanda za'a iya haɗa shi da ƙarfi ga kowane aikace-aikace.
Toarfin faɗaɗa terabytes na bayanai da ɗaruruwan biliyoyin maɓallan ajiya. - Ingantaccen aiki na dubunnan ayyuka masu daidaitawa.
- Increaseara mahimmanci a cikin bandwidth, rage latenci, da karatu / rubutu mafi girma don nau'ikan nau'ikan aiki idan aka kwatanta da wuraren aiki na yau da kullun.
- Ikon amfani da azuzuwan SSDs daban-daban a cikin ma'aji guda don inganta aiki da karko.
Kuna iya samun damar lambar injin daga mahaɗin da ke ƙasa.