kutulutsidwa kwa DBMS yotseguka , yopangidwa kuti isunge bwino magulu akuluakulu a data olumikizidwa omwe amapanga graph yomwe imatha kukhala ndi mabiliyoni a ma node ndi ma thililiyoni olumikizana. Ntchitoyi idalembedwa mu C ++ ndi zololedwa pansi pa Apache 2.0. Makasitomala amalaibulale ofikira ku DBMS amakonzekera zilankhulo za Go, Python ndi Java. Kuyambitsa chitukuko cha DBMS VESoft masiku angapo apitawo gawo loyamba la ndalama zokwana madola 8 miliyoni.
MU DBMS kugawa zomanga popanda kugawana zinthu (zogawana-palibe), kutanthauza kukhazikitsidwa kwa njira zodziyimira pawokha komanso zodziyimira pawokha zofunsira ma graph ndi njira zosungirako zosungidwa. Meta-service imayendetsa kayendetsedwe ka deta ndipo imapereka chidziwitso cha graph. Kuti mutsimikizire kusasinthika kwa data, algorithm-based protocol imagwiritsidwa ntchito .
Zofunikira zazikulu za Nebula Graph:
- Chitetezo popereka mwayi kwa ogwiritsa ntchito ovomerezeka okha omwe zilolezo zawo zimayikidwa kudzera mu dongosolo la Role-based Access Control (RBAC).
- kulumikiza mitundu yosiyanasiyana ya injini zosungira. Thandizo pakukulitsa chilankhulo chopanga mafunso ndi ma algorithms atsopano.
- Kuonetsetsa kuti mukuchedwa pang'ono powerenga kapena kulemba deta ndikusunga zambiri. Pa m'gulu la graphd node imodzi ndi ma database atatu osungidwa a 632 GB kukula kwake, kuphatikiza ma vertices 1.2 biliyoni ndi m'mphepete mabiliyoni 8.4, latencies anali pamlingo wa ma milliseconds angapo, ndipo zotsatira zake zinali zopempha 140 zikwi pa sekondi imodzi. .
- Linear scalability.
- Chilankhulo chofanana ndi SQL chomwe chili champhamvu komanso chosavuta kumva. Ntchito zothandizidwa zikuphatikiza GO (bidirectional traversal of graph vertices), GROUP BY, ORDER BY, LIMIT, UNION, UNION DISTINCT, INTERSECT, MINUS, PIPE (pogwiritsa ntchito zotsatira za funso lakale). Ma index ndi zosinthika zomwe zimafotokozedwa ndi ogwiritsa ntchito zimathandizidwa.
- Kuonetsetsa kupezeka kwakukulu ndi kupirira zolephera.
- Kuthandizira pakupanga zithunzithunzi ndi kagawo ka database kuti muchepetse kupanga makope osunga zobwezeretsera.
- Okonzeka kugwiritsidwa ntchito m'mafakitale (omwe amagwiritsidwa ntchito kale mu zomangamanga za JD, Meituan ndi Xiaohongshu).
- Kutha kusintha dongosolo losungiramo data ndikulisintha popanda kuyimitsa kapena kukhudza zomwe zikuchitika.
- Thandizo la TTL kuti muchepetse moyo wa data.
- Malamulo oyang'anira makonda ndi makamu osungira.
- Zida zoyendetsera ntchito ndikukonzekera kukhazikitsidwa kwa ntchito (zantchito zomwe zikuthandizidwa pano ndi COMPACT ndi FLUSH).
- Ntchito zopezera njira yathunthu ndi njira yayifupi kwambiri pakati pa ma vertices operekedwa.
- Mawonekedwe a OLAP ophatikizika ndi nsanja za gulu lachitatu.
- Zothandizira pakulowetsa deta kuchokera kumafayilo a CSV kapena kuchokera ku Spark.
- Tumizani ma metric owunikira pogwiritsa ntchito Prometheus ndi Grafana.
- Mawonekedwe a intaneti
pakuwona momwe ma graph amagwirira ntchito, kuyang'ana kwa ma graph, kupanga kusungirako deta ndi kutsitsa masinthidwe.
Source: opennet.ru