图数据库应用案例解析常新宇

QCon大会

2018/10/19 发布于 技术 分类

文字内容
1. ࢶහഝପଫአໜֺᥴຉ ଉෛਜ TigerGraph ਮಁᥴ٬ොໜᕪቘ
3. • य़හഝ෸դጱᳯ᷌ • ࢶහഝପጱս۠ • TigerGraphՕᕨ • አֺ1 —— ‫ݍ ᦪྂݍ‬။᰸ • አֺ2 —— चᏐᦡෞᓕቘ • አֺ3 —— ӱ‫ړۓ‬ຉ҅ࠟߝവគ • አֺ4 —— Ծӱ᱾ᓕቘ • ௛ᕮ Q&A
4. य़හഝ෸դጱᳯ᷌

5. ӡᇔ԰ᘶጱ෸դ • හഝጱ೮ᖅӧෙኞ౮ • ᐒԻᗑᕶਫ෸ࣁᕚ • ᇔᘶᗑ7*24೮ᖅᬩᤈ • ̶̶̶ ӥӞӻᳯ᷌? ᧆහইഝ֜‫ى‬೵ᘶയᙧහ‫ݸ‬ഝᵌጱᡐհԧ꧊ՋҘԍҘ
8. ࢶහഝପጱս۠

9. ࢶཛྷࣳ – ൈᬿӮኴ‫ى‬ᘶ๋ᛔᆐጱཛྷࣳ • හഝਂ‫ؙ‬ጱ๋ᛔᆐጱᤒᬡཛྷࣳ • Ꭳᦩࢶᨏ/୽ߥ‫ێ‬/๢࢏਍ԟ/… ๋ᛔᆐጱᦇᓒཛྷࣳ • ‫ݝ‬᭭ܲ஠ᥝጱහഝ
11. /0 1. /…aijSae”+"$ijĢÌn›d
 2. mP‡^ijħÄĎDŠ»¥Ėk`d
 3. /ij ·^Ħij (Ą
 4. /J0ijPageRankij×´$-ijÂfd> 10
12. TigerGraphՕᕨ

13. Ac/nf# 
 r- Œ dèØÈz ● ³ bóãm=yLdq§ ● ĢÌ óã„ÏÖ 
 „ÏÈz ● õĞFµ÷Ĝ;
 TigerGraph S¢ë«Y¯‘ }¬ J\g ● bm=d ● C++č%óãm=yLÿė ● GSQL¥Ėž¶ ● GraphStudiowˆSDK 12
14. O/Acƒ • GjH[TB…„ÏĶ8§6Û8100-200GBH[©c • íäjdZ=3 10+2dåÍĶ8§6Û8äZ=1Ī2ç+Ĵu åÍ • ƒ)GSQL— )į Qéd¥Ėž¶ijGj5¤}¬µ÷s½ \g 13
15. ᒫӣդࢶහഝପ ਫ෸ጱ୩‫۝‬௔ᚆ ੒ԭ᥶݊‫܉ک‬ӡᕆᜓᅩ޾ᬟጱັ ᧃӧ‫ک‬1ᑁᰦ Ԫ‫ࢶࣳۓ‬හഝପ ҁी‫ڢ‬දັ҂ ྯᑁ‫܈‬ӡग़ጱๅෛ҅ྯॠ‫܈‬ ՊᕆጱԪ‫ۓ‬ ṛ‫ݢ‬ಘ઀௔੒ԭ૤य़හഝᵞጱඪ೮ ‫܉‬Պӻᜓᅩ҅ӞӡՊ๵ᬟ Ⴎଶग़ྍ‫ړ‬ຉ 3ྍ‫ک‬10ग़ྍጱႮଶ‫ى‬ᔮັᧃଚӬᬰ ᤈ॔๥ጱᦇᓒ භఽහഝᵌᐺ‫כ‬ಷ चԭአಁጱ᥯ᜋ̵᮱ᳪ౲‫ݪل‬ᕟᕢഴ ‫ګ‬੒භఽහഝጱᦢᳯ๦ᴴ ฃԭ୏‫޾ݎ‬᮱ᗟ ᫷ຂֵአັᧃ᧍᥺ҁGSQL҂୏‫ݎ‬ଚӬአֵ አRESTfulള‫ݗ‬᫷ຂ᮱ᗟ॔๥ጱ‫ړ‬ຉ 14
16. 7uDLQ ‫ق‬ቖᒫӞጱࣁᕚඪ՞‫ྯ҅ݪل‬ॠ Պᕆၚ᪋አಁ ᗦࢵ๋य़ጱඪ՞๐‫ݪلۓ‬ ‫ق‬ቖਫᴬኞԾӾ๋य़ጱਫ෸Իฃࢶ ҁ100B+ӻᶮᅩ҅2B+ེྯॠਫ෸ๅෛ҂ ӱ‫ׁۓ‬ᩢࢶ ‫ق‬ቖ๋य़ᑏۖᐒԻኪৼࠟ‫ ݪلۓ‬ኪࠟԾߝᎣᦩࢶᨏ ‫ق‬ቖ๋य़‫و‬Ձԙ᫣ଫአ ‫ق‬ቖ๋य़ጱኪᗑ‫ݪل‬ ෭๜ᒫӞጱኪ‫ݪلמ‬ ᷚഴ҅‫ݳ‬ᥢ ኪ‫ێ‬ᗑ ᗑᕶਞ‫ق‬ 15
17. TigerGraph %V*„Ïġ „ÏËWO&P Loading Time 2-Hop Path Quer 16
18. TigerGraph %V*„Ïġ jŃ)&P Loading Time 2-Hop Path Quer 17
19. TigerGraph %V*„Ïġ ¥ĖO&P Loading Time 2-Hop Path Quer 18
20. TigerGraph %V*„Ïġ ¥ĖO&P Loading Time 2-Hop Path Quer 19
21. TigerGraph %V*„Ïġ ¥ĖO&P Loading Time 2-Hop Path Quer 20
22. GraphStudio ™ }¬9ü'$ UI Loading Time 2-Hop Path Quer 21
23. GSQL diQ«¥Ėž¶ Loading Time ᩙԨय़ռਫ෸ጱࢶ‫ړ‬ຉᚆ‫ • ێ‬चԭSQLጱ᧍ဩᦡᦇ • ෫ᦞฎฎහഝପᬡՈᬮฎෛಋ᮷‫ݢ‬զ᫷ຂӤಋ • ັᧃ‫ݢ‬զ᧣አৼັᧃ
 • ᔄ֒ԭ SQL ጱਂ‫ؙ‬ᬦᑕ • ࢶᶾਠෆ
 • ‫ݢ‬ฎਫሿձ֜ᓒဩ • ᛔۖਫሿଚᤈ॒ቘ ᛔۖ‫ۖސ‬TigerGraphጱय़ᥢཛྷଚᤈᦇᓒ୚ක चԭMapReduceཛྷࣳጱහഝ॒ቘ 2-Hop Path Quer 22
24. አֺ1 —— ‫ݍ ᦪྂݍ‬။᰸

25. ÃB7€à¿ - fĜIċŽdT’{ĕ ĐÁB:ęĬ= 7‚˜ ĝĈ‚˜ Stable group Empty stable group /// X X X 3-step friend relation Many in-group connections GGooooddPPhhoonneeFFeeaattuurreess (1) High call back phone (2) Stable group (3) Long term phone (4) Many in-group connections (5) 3-step friend relation Average distance > 3 Many rejected calls Bad Phone Features (1) Short term call duration (2) Empty stable group (3) No call back phone (4) Many rejected calls (5) Average distance > 3 24
26. ÃB7€à¿ - (îă„ÏZ=6Û.ij!ĭÁB:ęĬ • 8ädĜ4.6Īp,6į100Ī2ݍ% Æ2ݍ
 82&8p,6ɼ118Ò{ĕ‡OZ=6Û. Phone 1 Features (1) High call back phone (2) Stable group (3) Long term phone (4) Many in-group connections (5) 3-step friend relation Phone 2 Features (1) Short term call duration (2) Empty stable group (3) No call back phone (4) Many rejected calls (5) Avg. distance > 3 Training Data Tens – Hundreds of Billions of calls 25 ๢࢏਍ԟ
27. 26
28. አֺ2 —— चᏐᦡෞᓕቘ

29. %{ITh†U€;oY y9"xdI‹¸ěµ÷įøÂÙĥ ¯ēijÞ`Ԏ eĤ"xê >Sv‹ • Kb$ t]?kVp)Hij~ğČÀį ġįLUNįĆģ6Ûįr½Ûįn)ij J ĉsvh†Ċ{Er½ • <'5]%(IoT) =i@ij0n lL đÅįr½Û%Iċ¯²7‚ij!F$ -ĂáÞ` • Ywlj+;oijĚ4ĉsr½ LUNįr½ÛŽ‹n)ijmĩ[Ăáö[ •  ,s‚_~ˆ8ZiFE%{4 +ijÙClL eĴUeZ=<1Ģ y9BX‹Óar½ÛįīÑÅđ!¹ IċÄâ(A%º6@İď¨Ăáö[æ Ôİ Array: Storage Array is the container of logical pools Pool: Each has a port on storage array device & it is the container of logical LUNs Array Pool Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes: PhysicalCapacity, PromisedCapacity, Alert (1:n) Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes: Pool PhysicalCapacity, PromisedCapacity, Aler (1:n) LUN: Logical Unit Number LUN LUN Server: Each server may be assigned with multiple LUNs Server Application: An application can run on multiple servers. Applications may depend on other applications. Application Service: Each service calls multiple applications. BU: A business unit uses and pays for a service. LUN Server LUN Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes: PhysicalCapacity, PromisedCapacity, Alert (n:1) Server Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes:'>Attributes: IsDecommissioned (n:m)'>n:m)'>n:m)'>n:m) Application Service (n:m)'>n:m)'>n:m)'>n:m) (n:m)'>n:m)'>n:m)'>n:m) BU 28
30. [8ZVp^TC :C[8ZM' 1. FÕ¾r½°ñ Äùā ı~ğĵp]ij ҚÞijöûij¦Mij ÊÜijr½Ûijùªij ­ˆ¼ìIJ 2. ¾r½ĘĀ + wˆij ”ú¾r½Ġ)‰) 3. FĜp]¹ “ ½ùª ᮱ᳪ ᶱፓᕟ ᶱፓᕟ ᨮᨱՈ ᨮᨱՈ ᨮᨱՈ Ծߝ ኞԾሾह Ծߝ ၥᦶሾह ๐‫࢏ۓ‬ ๐‫࢏ۓ‬ ๐‫࢏ۓ‬ ‫ى‬ᳮ ᴳᕆ ‫܋‬ᕆ ୌᦓ ୌᦓ ୌᦓ ୌᦓ
31. አֺ3 ——ӱ‫ړۓ‬ຉ҅ࠟߝവគ

32. q(\9C • y9"xdI‹L}patients%prescribers3@TÂij !t?HòMôþe • >Sv‹ • 1 R–@Kćclaim„Ï@NTOij prescribers!referral)] • fĜclaims`&prescriberseg‰mß;o • {EA#ÐĒX(LHM)ijGJ„3z6%(IDN). *Dij!tY-­œ…Úï • 8 A#ÐĒXK"òijTNW})prescriber Špatienteg • y9BX‹†•òMXĮðý%ôþe 31
33. q(\9C ‫܅‬ኞ 2018-01-20 ኩᦫ የՈ 2018-01-23 ኩᦫ ‫܅‬ኞ ‫܅‬ኞԏᳵጱവគ‫ى‬ᔮ ‫܅‬ኞ 0.17 ‫܅‬ኞ 0.72 ‫܅‬ኞ ‫܅‬ኞ 0.51 ‫܅‬ኞ 0.32 0.41 PageRank ‫ݎ‬ሿ‫܅‬ኞጱ୽ߥ‫ێ‬ ‫܅‬ኞ ‫܅‬ኞ የՈ ‫܅‬ኞ ‫܅‬ኞ የՈ የՈ ‫܅‬ኞ ‫܅‬ኞ ‫܅‬ኞ ‫܅‬ኞ የՈ ‫܅‬ኞ ‫܅‬ኞ የՈ ‫܅‬ኞ ᐒ‫ݎ܄‬ሿᓒဩ‫ݎ‬ሿਫ֛ᵞᗭ
34. Q360 ਫ෸വគࢶ=
 ਫ෸ᶶਮ360ଶ + ਫ෸Ծߝࢶ ‫ق‬ො֖ࢶ Person B B likes hobby H Knows/likes/follows B Shops at Store Y B commented (to C) Is located at Z Item X has tag/ feature F Bought item X online from Store Y Item X is stored at Warehouse W Person A Would you like…? Person C C (also) shops at store Y Customer Visit Click Path
35. Q360 ં௔ ᘸᔄ ፘ֒ Ծߝ Ծߝ ᨻԣ አಁ ‫ى‬Ძਁ ፘ֒Ծߝവគ አಁᬮᨻԣԧᬯӻԾߝ አಁ ‫ݶ‬Ԫ҅Յ౵҅๏‫…݋‬ አಁ አಁ ં௔ አಁ വគ Ծߝ ᅩ‫҅ڋ‬තᡐ҅൤ᔱ҅ᨻԣ… Ծߝ አಁ चԭአಁᤈԅ҅ં௔҅አಁ‫ى‬ᔮ ጱԾߝവគ
36. አֺ4 —— Ծӱ᱾ᓕቘ

37. ᇔၞ޾‫׀‬ଫ᱾ᳯ᷌ਫᴬӤฎࢶᳯ᷌ ▪ ᇔၞᬩᬌӾ᫨ᒊᳵၿ݊ग़ྍࢶӤັᧃ ▪ ᇔාႴ‫ࣈ҅ܔ‬ቘ‫௳מ‬ၿ݊ग़੶ᕆᕮ຅ ▪ ኞԾ҅ପਂ҅ᬩᬌහഝᰁଲय़ ▪ ࢶ‫ݢ‬զ੒ኞԾࠟ҅‫׀‬ଫࠟ҅ᬩᬌࠟӨᦈ‫҅ܔ‬ ኞԾ҅ᨵᬩԏᳵ‫ى‬ᔮጱਠᗦ઀ሿ
38. ᇔၞ޾‫׀‬ଫ᱾ ࢶଫአਫֺ ᇔၞ޾‫׀‬ଫ᱾ॠᆐᭇ‫ړࢶݳ‬ຉ • Ծߝ᮱կׁᩢࢶ • ‫׀‬ଫ᱾‫ى‬ᔮࢶ • ࣈቘ֖ᗝࢶ • ᦈᨵ/ପਂ/հ໒‫ݒ‬۸॔๥‫ى‬ᘶࢶ … ࢶ‫ړ‬ຉਫֺ • ਫ෸ᭆᨵս۸ • ପਂ/ᦈ‫ړܔ‬ຉ • ᵭ᮱կհ໒/ପਂհ꧊୽ߥ‫ړ‬ຉ • մӱ‫ى‬ᘶ‫ړ‬ຉ
39. ᇔၞ޾‫׀‬ଫ᱾ ࢶଫአਫֺ Ծӱ᱾Ӿጱࢶአֺғ 1. ॓ᭌᬩᬌ᪠ᕚവគ 2. ᅒਸ਼୽ߥ‫ړ‬ຉ 3. ᇔාႴ‫܄ࣈ݊ᧃັܔ‬ग़੶ᕆັᧃ 4. ኞԾၞᑕഴ‫ګ‬᪙᪵
40. .2 • &CTÂO„Ïijd„Ï /R¡9 ijd„Ïġn )!†5‘—Oİ
 • fbdÅđÎ ijH -„ÏyLq§­ˆ¯‘L£İ TigerGraph ÇFOŒcąod„Ïq§İ
 • ®ęĬį®±į¿MK_!¹¦są>iĔdŸĨn)İ