Iyo nguva yekubika muExcel iri kukurumidza kunyangarika - maitiro akananga kune zviri nyore maturusi ekuratidza uye kuongorora ruzivo anoonekwa munzvimbo dzese. Isu tanga tichikurukura mukati medhijitari yekubika kwenguva yakareba uye takasarudza iyo Tableau yekuona uye yekuzvishandira analytics system. Alexander Bezugly, mukuru we analytical solutions uye dhipatimendi rekubika reM.Video-Eldorado Group, akataura pamusoro pezvakaitika nemigumisiro yekuvaka dashboard yehondo.
Ini ndichataura ipapo ipapo kuti hazvisi zvese zvakarongwa zvakazadzikiswa, asi chiitiko chainakidza, ndinovimba chichava chinobatsira kwauri zvakare. Uye kana paine munhu ane chero mazano ekuti zvingaitwe zvirinani sei, ndingatenda zvikuru nezano renyu uye mazano.
Pazasi pekucheka pane zvatakasangana nazvo uye zvatakadzidza nezvazvo.
Takatangira papi?
M.Vhidhiyo-Eldorado ine yakanyatso kugadziridzwa data modhi: yakarongeka ruzivo ine hudzamu hunodiwa hwekuchengetera uye nhamba huru yeyakagadziriswa-fomu mishumo (ona zvimwe zvakawanda ) Kubva pane izvi, vanoongorora vanogadzira matafura epivot kana mapepanhau akafomatirwa muExcel, kana yakanaka PowerPoint mharidzo yevashandisi vekupedzisira.
Anenge makore maviri apfuura, pachinzvimbo chemishumo yakamisikidzwa, takatanga kugadzira mishumo yekuongorora muSAP Analysis (iyo Excel yekuwedzera, chaizvo tafura yepivot pamusoro peinjini yeOLAP). Asi chishandiso ichi hachina kukwanisa kuzadzisa zvidikanwi zvevashandisi vese; ruzhinji rwakaramba rwuchishandisa ruzivo rwakawedzerwa nevanoongorora.
Vashandisi vedu vekupedzisira vanowira mumapoka matatu:
Top management. Inokumbira ruzivo nenzira yakanyatsoratidzwa uye inonzwisisika.
Middle management, vashandisi vepamusoro. Kufarira kuongorora data uye kukwanisa kuzvimiririra kuvaka mishumo kana maturusi aripo. Vakava vashandisi vanokosha vezvinyorwa zvekuongorora muSAP Analysis.
Vashandisi vakawanda. Ivo havafarire kuongorora data vakazvimirira; vanoshandisa mishumo ine rusununguko rushoma, muchimiro chezvinyorwa zvenhau uye matafura epivot muExcel.
Pfungwa yedu yaive yekuvhara zvido zvevashandisi vese uye kuvapa imwechete, yakanakira chishandiso. Takasarudza kutanga nevakuru vepamusoro. Vaida madhibhodhi ari nyore kushandisa kuongorora mhedzisiro yebhizinesi. Saka, takatanga neTableau uye takatanga kusarudza nzira mbiri: zviratidzo zvekutengesa uye zvekutengesa zvepamhepo zvine hudzamu hushoma uye hupamhi hwekuongorora, iyo yaizovhara inenge 80% yedata yakakumbirwa nevakuru vepamusoro.
Sezvo vashandisi vemadhibhodhi vaive manejimendi epamusoro, imwezve KPI yechigadzirwa yakaonekwa - kukurumidza kwekupindura. Hapana anomirira 20-30 masekondi kuti data rigadziriswe. Kufamba kunofanirwa kunge kwaitwa mukati me4-5 masekonzi, kana zvirinani, zvaitwa ipapo ipapo. Uye isu, nhamo, takakundikana kuita izvi.
Izvi ndizvo zvakaita marongerwo edashibhodhi redu guru:
Pfungwa yakakosha ndeyekubatanidza madhiraivha makuru eKPI, ayo aive gumi nemapfumbamwe akazara, kuruboshwe uye kuratidza masimba avo uye kuparara nehunhu hukuru kurudyi. Basa racho rinoita serakapfava, kuona kwacho kune musoro uye kunonzwisisika, kudzamara wanyura muzvinyorwa.
Detail 1. Data volume
Tafura yedu huru yekutengesa kwegore negore inotora mamiriyoni mazana matatu emitsara. Sezvo zvichidikanwa kuratidza masimba egore rapfuura uye gore rakapfuura, huwandu hwe data pane chaiyo yekutengesa chete inenge bhiriyoni mitsara. Ruzivo rwe data rakarongwa uye online yekutengesa block inochengetwawo zvakasiyana. Nokudaro, kunyange zvazvo takashandisa columnar in-memory DB SAP HANA, kukurumidza kwemubvunzo nekusarudzwa kwezviratidzo zvose kwevhiki imwe kubva pakuchengetedza ikozvino pane nhunzi yaiva inenge 300-1 seconds. Mhinduro yedambudziko iri inozviratidza pachayo - yekuwedzera materialization yedata. Asi zvakare ine misungo, zvimwe nezvavo pazasi.
Detail 2. Non-additive zviratidzo
Mazhinji eKPI edu akasungirirwa kune nhamba yemarisiti. Uye chiratidzo ichi chinomiririra COUNT DISTINCT yehuwandu hwemitsara (tarisa misoro) uye inoratidza huwandu hwakasiyana zvichienderana nehunhu hwakasarudzwa. Semuenzaniso, kuti chiratidzo ichi uye chinobva pairi chinofanira kuverengerwa sei:
Kuita kuti zviverengero zvako zvive zvechokwadi, unogona:
- Verenga zviratidzo zvakadaro panhunzi mukuchengetedza;
- Ita zviverengero pahuwandu hwe data muTableau, i.e. paunokumbira muTableau, ipa data rese zvinoenderana nemafirita akasarudzwa mune granularity yenzvimbo yekugamuchira;
- Gadzira chiratidziro chenyama umo zviratidzo zvese zvichaverengerwa mune ese masampuli sarudzo dzinopa akasiyana asiri ekuwedzera mhinduro.
Zviripachena kuti mumuenzaniso UTE1 uye UTE2 zvinhu zvenyama zvinomiririra iyo chigadzirwa hierarchy. Ichi hachisi chinhu chakamira; manejimendi mukati mekambani inoitika kuburikidza nazvo, nekuti Mamaneja akasiyana ane basa remapoka akasiyana ezvigadzirwa. Takava nedzokororo dzakawanda dzepasi rose dzehumambo uhwu, apo mazinga ese akachinja, apo hukama hwakadzokororwa, uye nguva dzose kuchinja kwenzvimbo, apo rimwe boka rakasimuka kubva pane imwe node kuenda kune imwe. Mukutaura kwakajairwa, zvese izvi zvinoverengerwa panhunzi kubva kuhunhu hwechinyorwa; kana iri nyaya yekugadzirwa kweiyi data, zvinodikanwa kugadzira nzira yekutevera shanduko dzakadai uye kurodha zvekare data renhoroondo. Basa risiri risingakoshi.
Detail 3. Data kuenzanisa
Pfungwa iyi yakafanana neyakapfuura. Chinokosha ndechekuti kana uchiongorora kambani, itsika kuumba akati wandei mazinga ekuenzanisa nenguva yapfuura:
Kuenzanisa nenguva yapfuura (zuva nezuva, vhiki nevhiki, mwedzi nemwedzi)
Mukuenzanisa uku, zvinofungidzirwa kuti zvichienderana nenguva yakasarudzwa nemushandisi (somuenzaniso, vhiki yemakumi matatu nenhatu yegore), tinofanira kuratidza masimba nevhiki ye33; kana takasarudza data kwemwedzi, semuenzaniso, Chivabvu. , zvino kuenzanisa uku kwaizoratidza masimba panosvika Kubvumbi.
Kuenzanisa negore rapfuura
Iyo huru nuance pano ndeyokuti kana uchienzanisa nezuva uye nevhiki, iwe hausi kutora zuva rimwechete regore rapfuura, i.e. haugone kungoisa gore razvino bvisa rimwechete. Unofanira kutarisa zuva revhiki rauri kuenzanisa. Kana uchienzanisa mwedzi, pane zvinopesana, iwe unofanirwa kutora chaiwo iwo chaiwo kalendari yegore rapfuura. Kune zvakare nuances ane leap years. Muzvinyorwa zvepakutanga, ruzivo rwese rwunogoverwa masikati; hapana minda yakaparadzana ine mavhiki, mwedzi, kana makore. Naizvozvo, kuti uwane yakakwana analytical cross-section mupaneri, haufanirwe kuverenga kwete imwe nguva, semuenzaniso vhiki, asi 4 mavhiki, uye wozoenzanisa data idzi, ratidza masimba, kutsauka. Saizvozvo, iyi pfungwa yekugadzira kuenzanisa mune dynamics inogona zvakare kuitwa ingave muTableau kana kudivi rechitoro. Hongu, uye zvechokwadi isu taiziva uye takafunga nezvemashoko aya panguva yekugadzira, asi zvakanga zvakaoma kufanotaura maitiro avo pakuita kwedashboard yekupedzisira.
Pakuita dashboard, takatevera nzira refu yeAgile. Basa redu raive rekupa chishandiso chekushanda chine data inodiwa yekuyedza nekukurumidza sezvinobvira. Naizvozvo, takaenda muma sprints uye takatanga kubva mukuderedza basa padivi rekuchengetedza ikozvino.
Chikamu 1: Kutenda muTableau
Kurerutsa rutsigiro rweIT uye nekukasira kuita shanduko, isu takasarudza kugadzira iyo pfungwa yekuverenga isiri-yekuwedzera zviratidzo uye kuenzanisa nguva dzakapfuura muTableau.
Nhanho 1. Zvese zviri Live, hapana gadziriso yehwindo.
Panguva ino, takabatanidza Tableau kune zvitoro zvazvino uye takasarudza kuona kuti nhamba yemarisiti egore rimwe yaizoverengwa sei.
Mhinduro:
Mhinduro yacho yaiodza mwoyo - maminitsi makumi maviri. Kuchinjisa data pamusoro petiweki, mutoro wakakwira paTableau. Takaona kuti logic ine zvisiri-additive zviratidzo zvinoda kuitwa paHANA. Izvi hazvina kutityisa zvakanyanya, takanga tatova neruzivo rwakafanana neBO uye Analysis uye taiziva nzira yekuvaka mashowcases anokurumidza muHANA ayo anoburitsa zvakaverengerwa zvisiri zvekuwedzera zviratidzo. Zvino chakanga chasara kuvagadzirisa kuti vaende kuTable.
Nhanho 2. Isu tinogadzirisa zviratidziro, hapana zvinhu, zvese zviri pamhepo.
Isu takagadzira yakaparadzana nyowani showcase yakagadzira iyo inodiwa data yeTABLEAU panhunzi. Kazhinji, isu takawana mhedzisiro yakanaka; isu takadzikisa nguva yekugadzira zvese zviratidzo muvhiki imwe kusvika 9-10 masekonzi. Uye isu takatarisira nemoyo wese kuti muTableau nguva yekupindura yedashboard ingave 20-30 seconds pakuvhurwa kwekutanga uyezve nekuda kwecache kubva pa10 kusvika ku12, izvo zvakajairika isu.
Mhinduro:
Dashboard yekutanga yakavhurika: 4-5 maminitsi
Chero kubaya: 3-4 maminitsi
Hapana aitarisira kuwedzera kwakadai kwekuwedzera kwebasa repamberi pechitoro.
Chikamu 2. Dive muTableau
Nhanho 1. Tableau performance analysis uye kukurumidza kugadzirisa
Takatanga kuongorora kuti Tafura inopedzera nguva yakawanda papi. Uye kune maturusi akanaka eizvi, ayo, ehe, ari kuwedzera kweTableau. Dambudziko guru ratakawona raive mibvunzo yakaoma kwazvo yeSQL yaivaka Tableau. Ivo vainyanya kubatanidzwa ne:
- kushandura data. Sezvo Tableau isina maturusi ekufambisa dhatabheti, kuvaka rutivi rworuboshwe rwedashibhodhi rine ratidziro yakadzama yemaKPI ese, taifanira kugadzira tafura tichishandisa kesi. Saizi yeSQL mibvunzo mudhatabhesi yakasvika 120 mavara.
- sarudzo yenguva yenguva. Mubvunzo wakadaro padanho redatabase wakatora nguva yakawanda kuunganidza pane kuita:
Avo. kukumbira kugadzirisa 12 seconds + 5 seconds execution.
Isu takasarudza kurerutsa iyo yekuverenga logic padivi reTableau uye nekufambisa imwe chikamu chekuverenga kune kumberi kwechitoro uye dhatabhesi level. Izvi zvakaunza migumisiro yakanaka.
Kutanga, isu takaita shanduko pane nhunzi, takazviita kuburikidza nekubatanidzwa kwekunze kwekunze padanho rekupedzisira rekuverenga VIEW, maererano nenzira iyi inotsanangurwa pawiki. ΠΈ .
Ndiko kuti, takagadzira tafura yekuisa - transposition matrix (21x21) uye takagamuchira zviratidzo zvose mumutsara-ne-mutsara kuputsika.
Waive:
Yakava:
Inenge hapana nguva inopedzwa pane dhatabhesi transposition pachayo. Chikumbiro chezviratidzo zvose zvevhiki yakaramba ichigadziriswa mukati memasekondi gumi. Asi kune rumwe rutivi, kushanduka kwakarasika maererano nekuvaka dashboard zvichienderana nechiratidzo chaicho, i.e. kudivi rekurudyi redhibhodhi uko kusimba uye kudzika kwakadzama kwechiratidzo chaicho chinoratidzwa, yaimbove iyo kesi yekuratidzira yakashanda mumasekonzi 10-1, nekuti. chikumbiro chaive chakavakirwa pane chimwe chiratidzo, uye ikozvino dhatabhesi yaigara ichisarudza zviratidzo zvese uye yakasefa mhedzisiro isati yadzosera mhedzisiro kuTableau.
Nekuda kweizvozvo, kumhanya kwedashboard kwakadzikira kanenge ka3.
Mhinduro:
- 5 sec - kupatsanura madhibhodhi, zviono
- 15-20 masekondi - kugadzirira kuunganidza mibvunzo nekuita pre-maverengero muTableau
- 35-45 sec - kuunganidzwa kwemibvunzo yeSQL uye yavo yakafanana-yakatevedzana kuuraya muHana
- 5 sec - kugadzirisa mhedzisiro, kuronga, kuverengera zviono muTableau
- Ehe, mhedzisiro yakadaro haina kuenderana nebhizinesi, uye isu takaenderera mberi nekugadzirisa.
Nhanho 2. Minimum logic muTableau, kukwana zvinhu
Isu takanzwisisa kuti zvaisaita kuvaka dashboard ine nguva yekupindura yemasekonzi akati wandei pane chitoro chinomhanya kwemasekonzi gumi, uye isu takafunga nezvesarudzo dzekugadzira data padivi redatabase zvakananga kune dashboard inodiwa. Asi isu takasangana nedambudziko repasi rose rakatsanangurwa pamusoro - zvisiri zvekuwedzera zviratidzo. Hatina kukwanisa kuve nechokwadi chekuti pakuchinja mafirita kana madhiri, Tableau inochinja-chinja pakati pezvitoro zvakasiyana uye mazinga akafanogadzirirwa kune akasiyana echigadzirwa hierarchies (mumuenzaniso, mibvunzo mitatu isina UTE, ine UTE10 uye UTE1 inoburitsa mhedzisiro dzakasiyana). Naizvozvo, isu takasarudza kurerutsa dashboard, kusiya iyo chigadzirwa hierarchy mudhibhodhi uye toona kuti ingaite kukurumidza sei mushanduro yakareruka.
Saka, pane ino nhanho yekupedzisira, isu takaunganidza yakaparadzana repository umo isu takawedzera ese maKPIs mune transposed fomu. Padivi re database, chero chikumbiro chekuchengetedza kwakadaro chinogadziriswa mu 0,1 - 0,3 masekondi. Mudhishibhodhi takagamuchira zvinotevera:
Kutanga kuvhura: 8-10 masekondi
Chero kudzvanya: 6-7 masekonzi
Nguva inoshandiswa neTableau inosanganisira:
- 0,3 sec. -Dashboard parsing uye kuunganidzwa kweSQL mibvunzo
- 1,5-3 sec. -Kuitwa kweSQL mibvunzo muHana yezviratidziro zvikuru (inomhanya ichienderana nedanho 1)
- 1,5-2 sec. - kupa, kuverengazve kwezviono
- 1,3sec. -Kuitwa kwemamwe mibvunzo yeSQL kuti uwane akakodzera mafirita emhando (Brand, Division, Guta, Chitoro), parsing mhedzisiro
Kuzvipfupikisa muchidimbu
Isu takafarira chishandiso cheTableau kubva pakuona kwekuona. Pachikamu cheprototyping, takafunga zvakasiyana-siyana zvekuona zvinhu uye tikazviwana zvese mumaraibhurari, kusanganisira yakaoma-akawanda-nhanho segmentation uye akawanda-mutyairi mapopoma emvura.
Tichiri kushandisa dashboards ane akakosha ekutengesa zviratidzo, takasangana nematambudziko ekuita ayo isu tisati takwanisa kukunda. Takapedza mwedzi inopfuura miviri uye takagamuchira dashboard isina kukwana, kukurumidza kwekupindura iyo iri pedyo nekugamuchirwa. Uye takazvisarudzira isu pachedu:
- Tafura haigone kushanda nehuwandu hwakawanda hwe data. Kana mune yepakutanga data modhi une anopfuura 10 GB yedata (inenge 200 miriyoni X 50 mitsara), ipapo dashboard inodzika zvakanyanya - kubva pamasekonzi gumi kusvika kumaminitsi akati wandei pakudzvanya kwega kwega. Takaedza nezvose zviri zviviri live-connect uye extract. Kumhanya kwekushanda kunofananidzwa.
- Kuganhurirwa kana uchishandisa akawanda matura (datasets). Iko hakuna nzira yekuratidza hukama pakati pe dataset uchishandisa yakajairwa nzira. Kana iwe ukashandisa workaround kubatanidza dataset, izvi zvinokanganisa zvakanyanya kuita. Kwatiri isu, takafunga nezve sarudzo yekugadzira data mune yega yega inodiwa yekuona chikamu uye nekuchinja pane aya akabatika dataset tichichengetedza mafirita akambosarudzwa - izvi zvakazove zvisingaite muTableau.
- Izvo hazvigoneke kuita dynamic parameters muTableau. Iwe haugone kuzadza paramita inoshandiswa kusefa dhatabheti mune yakatorwa kana panguva yekurarama-batanidza nemhedzisiro yeimwe sarudzo kubva kudhatabheti kana mhedzisiro yeimwe SQL query, chete chizvarwa chekuisa mushandisi kana chinogara chiripo.
- Zvipimo zvine chekuita nekuvaka dashboard ine OLAP | PivotTable zvinhu.
MuMSTR, SAP SAC, SAP Analysis, kana iwe ukawedzera dheti kune mushumo, ipapo zvinhu zvose zviri pairi zvakabatana nekusavapo. Tableau haina izvi; iyo yekubatanidza inofanirwa kugadzirwa nemaoko. Izvi zvimwe zvinogona kuchinjika, asi kune edu ese madhibhodhi ichi chinhu chinosungirwa kuzvinhu - saka iyi ndeimwe mari yevashandi. Uyezve, kana iwe ukaita mafirita ane hukama kuitira kuti, semuenzaniso, paunenge uchisefa dunhu, rondedzero yemaguta inongogumira kumaguta edunhu rino, iwe unobva wangopedzisira wava nemibvunzo inotevedzana kune dhatabhesi kana Extract, iyo inononoka kunonoka. dashboard. - Kuganhurirwa mumabasa. Shanduko yemisa haigone kuitwa pane zvakatorwa kana, ZVIKURU, pane dataset kubva kuLive-connecta. Izvi zvinogona kuitwa kuburikidza neTableau Prep, asi ibasa rekuwedzera uye chimwe chishandiso chekudzidza nekuchengetedza. Semuenzaniso, haugone kufambisa data kana kuibatanidza nayo pachayo. Chii chakavharwa kuburikidza neshanduko pamakoramu ega ega kana minda, iyo inofanirwa kusarudzwa kuburikidza nekesi kana kana, uye izvi zvinogadzira yakaoma kwazvo SQL mibvunzo, umo iyo dhatabhesi inopedza yakawanda yenguva yayo kuunganidza query zvinyorwa. Uku kusachinjika kwechishandiso kwaifanira kugadziriswa padanho rekuratidzira, izvo zvinotungamira kune yakanyanya kuoma kuchengetedza, kuwedzera kurodha uye shanduko.
Hatina kurasa Tafura. Asi isu hatitarise Tableau sechishandiso chinokwanisa kuvaka madhibhodhi eindasitiri uye chishandiso chekutsiva nekuita dhijitari yekambani yese yekushuma system.
Isu tave kushingairira kugadzira dashibhodhi rakafanana mune chimwe chishandiso uye, panguva imwe chete, tichiedza kudzokorora dhibhodhi yekuvaka muTableau kuitira kuti ive nyore zvakanyanya. Kana nharaunda ichifarira, tichakuudza pamusoro pemigumisiro.
Isu takamirirawo mazano ako kana zano rekuti muTabeau unogona sei kuvaka madhibhodhi ekukurumidza pamusoro peakawanda mavhoriyamu edata, nekuti isu tine webhusaiti uko kune data rakawanda kupfuura rekutengesa.
Source: www.habr.com