Mear sidestatistiken yn jo lytse opslach

Troch sidestatistiken te analysearjen krije wy in idee fan wat der mei bart. Wy fergelykje de resultaten mei oare kennis oer it produkt of tsjinst en ferbetterje dêrmei ús ûnderfining.

As de analyze fan 'e earste resultaten is foltôge, de ynformaasje is begrepen en konklúzjes binne lutsen, begjint de folgjende poadium. Ideeën ûntsteane: wat bart der as jo nei de gegevens fan 'e oare kant sjogge?

D'r binne beheiningen fan 'e analyse-ark op dit stadium. Dit is ien fan 'e redenen wêrom't Google Analytics my net genôch wie, nammentlik fanwegen de beheinde mooglikheid om myn gegevens te sjen en te manipulearjen.

Ik woe altyd basisgegevens (mastergegevens) fluch laden, in oar nivo fan aggregaasje tafoegje, of besteande wearden oars ynterpretearje.

Dit is maklik te dwaan yn yn jo lytse opslachromte basearre op de access.log triem en de SQL taal is genôch foar dit.

Dus, hokker fragen woe ik beäntwurde?

Wat en wannear feroare op de side

De skiednis fan feroaringen yn 'e ûnderlizzende gegevens (mastergegevens) is altyd fan belang.

SQL rapport query

SELECT
	1 as 'SideStackedBar: Content Updates by Months',
	strftime('%m/%Y', datetime(UPDATE_DT, 'unixepoch')) AS 'Day',
	COUNT(CASE WHEN PAGE_TITLE != 'n.a.' THEN DIM_REQUEST_ID END) AS 'Web page updates',
	COUNT(CASE WHEN PAGE_DESCR = 'IMAGES' THEN DIM_REQUEST_ID END) AS 'Image uploads',
	COUNT(CASE WHEN PAGE_DESCR = 'VIDEO' THEN DIM_REQUEST_ID END) AS 'Video uploads',
	COUNT(CASE WHEN PAGE_DESCR = 'AUDIO' THEN DIM_REQUEST_ID END) AS 'Audio uploads'
FROM DIM_REQUEST
WHERE PAGE_TITLE != 'n.a.' OR PAGE_DESCR != 'n.a.'
GROUP BY strftime('%m/%Y', datetime(UPDATE_DT, 'unixepoch'))
ORDER BY UPDATE_DT

Bygelyks, op in stuit waard sykmasino-optimisaasje útfierd of nije ynhâld waard tafoege oan 'e side, en dêrom wurdt in ferheging fan ferkear ferwachte.

Groep brûkers

It ienfâldichste foarbyld fan in groep is de brûkersagint of de namme fan it bestjoeringssysteem.

De diminsje fan brûkersagent hat sawat tûzen records sammele en ik wie ynteressearre yn it sjen fan de dynamyk fan 'e ferdieling fan aginten binnen de groep.

SQL rapport query

SELECT
	1 AS 'SideStackedBar: User Agents',
	AGENT_OS AS 'OS',
	SUM(CASE WHEN AGENT_BOT = 'n.a.' THEN 1 ELSE 0 END ) AS 'User Agent of Users',
	SUM(CASE WHEN AGENT_BOT != 'n.a.' THEN 1 ELSE 0 END ) AS 'User Agent of Bots'
FROM DIM_USER_AGENT
WHERE DIM_USER_AGENT_ID != -1
GROUP BY AGENT_OS
ORDER BY 3 DESC

De measte ferskillende kombinaasjes fan aginten komme nei de side fan 'e Windows-wrâld. Under de net identifisearre wiene WhatsApp, PocketImageCache, PlayStation, SmartTV, ensfh.

Aktiviteit fan brûkersgroep per wike

Troch guon groepen te kombinearjen, kin men de ferdieling fan har aktiviteit observearje.

Bygelyks, Linux-kluster brûkers konsumearje mear websideferkear dan elkenien oars.

SQL rapport query

SELECT
1 as 'StackedBar: Traffic Volume by User OS and by Week',
strftime('%W week', datetime(FCT.EVENT_DT, 'unixepoch')) AS 'Week',
SUM(CASE WHEN USG.AGENT_OS IN ('Android', 'Linux') THEN FCT.BYTES ELSE 0 END)/1000 AS 'Android/Linux Users',
SUM(CASE WHEN USG.AGENT_OS IN ('Windows') THEN FCT.BYTES ELSE 0 END)/1000 AS 'Windows Users',
SUM(CASE WHEN USG.AGENT_OS IN ('Macintosh', 'iOS') THEN FCT.BYTES ELSE 0 END)/1000 AS 'Mac/iOS Users',
SUM(CASE WHEN USG.AGENT_OS IN ('n.a.', 'BlackBerry') THEN FCT.BYTES ELSE 0 END)/1000 AS 'Other'
FROM
  FCT_ACCESS_USER_AGENT_DD FCT,
  DIM_USER_AGENT USG,
  DIM_HTTP_STATUS HST
WHERE FCT.DIM_USER_AGENT_ID=USG.DIM_USER_AGENT_ID
  AND FCT.DIM_HTTP_STATUS_ID = HST.DIM_HTTP_STATUS_ID
  AND USG.AGENT_BOT = 'n.a.' /* users only */
  AND HST.STATUS_GROUP IN ('Successful') /* good pages */
  AND datetime(FCT.EVENT_DT, 'unixepoch') > date('now', '-3 month')
GROUP BY strftime('%W week', datetime(FCT.EVENT_DT, 'unixepoch'))
ORDER BY FCT.EVENT_DT

Yntinsive ferkearskonsumpsje

De tabel lit de meast aktive brûkersgroepen sjen en de dei fan har aktiviteit.
De meast aktive hearre ta it Linux-kluster.

SQL rapport query

SELECT
1 AS 'Table: User Agent with Havy Usage',
strftime('%d.%m.%Y', datetime(FCT.EVENT_DT, 'unixepoch')) AS 'Day',
ROUND(1.0*SUM(FCT.BYTES)/1000000, 1) AS 'Traffic MB',
ROUND(1.0*SUM(FCT.IP_CNT)/SUM(1), 1) AS 'IPs',
ROUND(1.0*SUM(FCT.REQUEST_CNT)/SUM(1), 1) AS 'Requests',
USA.DIM_USER_AGENT_ID AS 'ID',
MAX(USA.USER_AGENT_NK) AS 'User Agent',
MAX(USA.AGENT_BOT) AS 'Bot'
FROM
FCT_ACCESS_USER_AGENT_DD FCT,
DIM_USER_AGENT USA
WHERE FCT.DIM_USER_AGENT_ID = USA.DIM_USER_AGENT_ID
  AND datetime(FCT.EVENT_DT, 'unixepoch') >= date('now', '-30 day')
GROUP BY USA.DIM_USER_AGENT_ID, strftime('%d.%m.%Y', datetime(FCT.EVENT_DT, 'unixepoch')) 
ORDER BY SUM(FCT.BYTES) DESC, FCT.EVENT_DT
LIMIT 10

Mei de dei- en agent-ID-attributen wurdt it mooglik statistiken fluch te finen en te folgjen oer de dagen fan yndividuele brûkersgroepen. As it nedich is, kinne jo fluch detaillearre ynformaasje fine yn 'e poadiumtabel.

Hoe kinne jo ynformaasje krije?

Ynformaasje út de access.log triem kin noch effisjinter makke wurde troch ekstra gegevensboarnen te yntegrearjen en nije nivo's fan aggregaasje en groepearring yn te fieren.

Basisgegevens en entiteiten

Basisgegevens omfetsje ynformaasje oer entiteiten: websiden, foto's, fideo- en audioynhâld, yn it gefal fan in winkel - produkten.

De entiteiten sels fungearje as mjittingen, en it proses fan it bewarjen fan feroaringen yn attributen wurdt histoarisearring neamd. Yn in databank wurdt dit proses faak útfierd yn 'e foarm fan stadich fariearjende dimensjes (SCD).

De boarne gegevens kinne komme út in ferskaat oan systemen, sadat se hast altyd moatte wurde yntegrearre.

Stadich feroarjende diminsje

De diminsje DIM_REQUEST sil ynformaasje befetsje oer oanfragen op 'e side yn histoaryske foarm.

Tabel SCD2

CREATE TABLE DIM_REQUEST ( /* scd table for user requests */
  DIM_REQUEST_ID      INTEGER NOT NULL PRIMARY KEY AUTOINCREMENT,
  DIM_REQUEST_ID_HIST INTEGER NOT NULL DEFAULT -1,
  REQUEST_NK          TEXT NOT NULL DEFAULT 'n.a.', /* request without ?parameters */
  PAGE_TITLE          TEXT NOT NULL DEFAULT 'n.a.',
  PAGE_DESCR          TEXT NOT NULL DEFAULT 'n.a.',
  PAGE_KEYWORDS       TEXT NOT NULL DEFAULT 'n.a.',
  DELETE_FLAG         INTEGER NOT NULL DEFAULT 0,
  UPDATE_DT           INTEGER NOT NULL DEFAULT 0,
  UNIQUE (REQUEST_NK, DIM_REQUEST_ID_HIST)
);
INSERT INTO DIM_REQUEST (DIM_REQUEST_ID) VALUES (-1);

Njonken it sille wy ien werjefte meitsje dy't altyd alle records yn 'e lêste steat toant. Needsaaklik foar it laden fan de mjitting sels.

Aktuele fertsjintwurdiging fan SCD2

/* Content: actual view on scd table */
SELECT HI.DIM_REQUEST_ID,
  HI.DIM_REQUEST_ID_HIST,
  HI.REQUEST_NK,
  HI.PAGE_TITLE,
  HI.PAGE_DESCR,
  HI.PAGE_KEYWORDS,
  NK.CNT AS HIST_CNT,
  HI.DELETE_FLAG,
  strftime('%d.%m.%Y %H:%M', datetime(HI.UPDATE_DT, 'unixepoch')) AS UPDATE_DT
FROM
  ( SELECT REQUEST_NK, MAX(DIM_REQUEST_ID) AS DIM_REQUEST_ID, SUM(1) AS CNT
    FROM DIM_REQUEST
    GROUP BY REQUEST_NK
  ) NK,
  DIM_REQUEST HI
WHERE 1 = 1
  AND NK.REQUEST_NK = HI.REQUEST_NK
  AND NK.DIM_REQUEST_ID = HI.DIM_REQUEST_ID;

En in werjefte wêr't histoaryske ynformaasje wurdt sammele foar elke yngong. It is needsaaklik om in histoarysk korrekte ferbining te bouwen mei de feiten.

Histoaryske presintaasje fan SCD2

/* Content: actual view on scd table */
SELECT SCD.DIM_REQUEST_ID,
  SCD.DIM_REQUEST_ID_HIST,
  SCD.REQUEST_NK,
  SCD.PAGE_TITLE,
  SCD.PAGE_DESCR,
  SCD.PAGE_KEYWORDS,
  SCD.DELETE_FLAG,
  CASE
    WHEN HIS.UPDATE_DT IS NULL
    THEN 1
    ELSE 0 END ACTIVE_FLAG,
  SCD.DIM_REQUEST_ID_HIST AS ID_FROM,
  SCD.DIM_REQUEST_ID AS ID_TO,
  CASE
    WHEN SCD.DIM_REQUEST_ID_HIST=-1
    THEN 3600
    ELSE IFNULL(SCD.UPDATE_DT,3600)
  END AS TIME_FROM,
  CASE
    WHEN HIS.UPDATE_DT IS NULL
    THEN 253370764800
    ELSE HIS.UPDATE_DT
  END AS TIME_TO,
  CASE
    WHEN SCD.DIM_REQUEST_ID_HIST=-1
    THEN STRFTIME('%d.%m.%Y %H:%M', DATETIME(3600, 'unixepoch'))
    ELSE STRFTIME('%d.%m.%Y %H:%M', DATETIME(IFNULL(SCD.UPDATE_DT,3600), 'unixepoch'))
  END AS ACTIVE_FROM,
  CASE
    WHEN HIS.UPDATE_DT IS NULL
    THEN STRFTIME('%d.%m.%Y %H:%M', DATETIME(253370764800, 'unixepoch'))
    ELSE STRFTIME('%d.%m.%Y %H:%M', DATETIME(HIS.UPDATE_DT, 'unixepoch'))
  END AS ACTIVE_TO
FROM
  DIM_REQUEST SCD
  LEFT OUTER JOIN DIM_REQUEST HIS
  ON SCD.REQUEST_NK = HIS.REQUEST_NK AND SCD.DIM_REQUEST_ID = HIS.DIM_REQUEST_ID_HIST;

Data Aggregaasje

Kompresje (aggregaasje) lit jo gegevens op in heger nivo evaluearje en anomalies en trends ûntdekke dy't net sichtber binne yn detaillearre rapporten.

Foegje bygelyks in groep ta oan de diminsje mei fersykstatuskoades DIM_HTTP_STATUS:

STATUS/GROUP
0xx/yn
1xx / Ynformaasje
2xx/Suksesfol
3xx/Omlieding
4xx/Client Flater
5xx / tsjinner flater

De diminsje fan brûkersagent DIM_USER_AGENT sil de attributen AGENT_OS en AGENT_BOT befetsje dy't ferantwurdlik binne foar groepen. Dizze kinne wurde befolke tidens it ETL-proses:

DIM_USER_AGENT wurdt laden

/* Propagate the user agent from access log */
INSERT INTO DIM_USER_AGENT (USER_AGENT_NK, AGENT_OS, AGENT_ENGINE, AGENT_DEVICE, AGENT_BOT, UPDATE_DT)
WITH CLS AS (
	SELECT BROWSER
	FROM STG_ACCESS_LOG WHERE LENGTH(BROWSER)>1
	GROUP BY BROWSER
)
SELECT
	CLS.BROWSER AS USER_AGENT_NK,
	CASE
	WHEN INSTR(CLS.BROWSER,'Macintosh')>0
		THEN 'Macintosh'
	WHEN INSTR(CLS.BROWSER,'iPhone')>0
			 OR INSTR(CLS.BROWSER,'iPad')>0
			 OR INSTR(CLS.BROWSER,'iPod')>0
			 OR INSTR(CLS.BROWSER,'Apple TV')>0
			 OR INSTR(CLS.BROWSER,'Darwin')>0
		THEN 'iOS'
	WHEN INSTR(CLS.BROWSER,'Android')>0
		THEN 'Android'
	WHEN INSTR(CLS.BROWSER,'X11;')>0 OR INSTR(CLS.BROWSER,'Wayland;')>0 OR INSTR(CLS.BROWSER,'linux-gnu')>0
		THEN 'Linux'
	WHEN INSTR(CLS.BROWSER,'BB10;')>0 OR INSTR(CLS.BROWSER,'BlackBerry')>0
		THEN 'BlackBerry'
	WHEN INSTR(CLS.BROWSER,'Windows')>0
		THEN 'Windows'
	ELSE 'n.a.' END AS AGENT_OS, -- OS
	CASE
	WHEN INSTR(CLS.BROWSER,'AppleCoreMedia')>0
		THEN 'AppleWebKit'
	WHEN INSTR(CLS.BROWSER,') ')>1 AND LENGTH(CLS.BROWSER)>INSTR(CLS.BROWSER,') ')
		THEN COALESCE(SUBSTR(CLS.BROWSER, INSTR(CLS.BROWSER,') ')+2, LENGTH(CLS.BROWSER) - INSTR(CLS.BROWSER,') ')-1), 'N/A')
	ELSE 'n.a.' END AS AGENT_ENGINE, -- Engine
	CASE
	WHEN INSTR(CLS.BROWSER,'iPhone')>0
		THEN 'iPhone'
	WHEN INSTR(CLS.BROWSER,'iPad')>0
		THEN 'iPad'
	WHEN INSTR(CLS.BROWSER,'iPod')>0
		THEN 'iPod'
	WHEN INSTR(CLS.BROWSER,'Apple TV')>0
		THEN 'Apple TV'
	WHEN INSTR(CLS.BROWSER,'Android ')>0 AND INSTR(CLS.BROWSER,'Build')>0
		THEN COALESCE(SUBSTR(CLS.BROWSER, INSTR(CLS.BROWSER,'Android '), INSTR(CLS.BROWSER,'Build')-INSTR(CLS.BROWSER,'Android ')), 'n.a.')
	WHEN INSTR(CLS.BROWSER,'Android ')>0 AND INSTR(CLS.BROWSER,'MIUI')>0
		THEN COALESCE(SUBSTR(CLS.BROWSER, INSTR(CLS.BROWSER,'Android '), INSTR(CLS.BROWSER,'MIUI')-INSTR(CLS.BROWSER,'Android ')), 'n.a.')
	ELSE 'n.a.' END AS AGENT_DEVICE, -- Device
	CASE
	WHEN INSTR(LOWER(CLS.BROWSER),'yandex.com')>0
		THEN 'yandex'
	WHEN INSTR(LOWER(CLS.BROWSER),'googlebot')>0
		THEN 'google'
	WHEN INSTR(LOWER(CLS.BROWSER),'bingbot')>0
		THEN 'microsoft'
	WHEN INSTR(LOWER(CLS.BROWSER),'ahrefsbot')>0
		THEN 'ahrefs'
	WHEN INSTR(LOWER(CLS.BROWSER),'jobboersebot')>0 OR INSTR(LOWER(CLS.BROWSER),'jobkicks')>0
		THEN 'job.de'
	WHEN INSTR(LOWER(CLS.BROWSER),'mail.ru')>0
		THEN 'mail.ru'
	WHEN INSTR(LOWER(CLS.BROWSER),'baiduspider')>0
		THEN 'baidu'
	WHEN INSTR(LOWER(CLS.BROWSER),'mj12bot')>0
		THEN 'majestic-12'
	WHEN INSTR(LOWER(CLS.BROWSER),'duckduckgo')>0
		THEN 'duckduckgo'
	WHEN INSTR(LOWER(CLS.BROWSER),'bytespider')>0
		THEN 'bytespider'
	WHEN INSTR(LOWER(CLS.BROWSER),'360spider')>0
		THEN 'so.360.cn'
	WHEN INSTR(LOWER(CLS.BROWSER),'compatible')>0 OR INSTR(LOWER(CLS.BROWSER),'http')>0
		OR INSTR(LOWER(CLS.BROWSER),'libwww')>0 OR INSTR(LOWER(CLS.BROWSER),'spider')>0
		OR INSTR(LOWER(CLS.BROWSER),'java')>0 OR INSTR(LOWER(CLS.BROWSER),'python')>0
		OR INSTR(LOWER(CLS.BROWSER),'robot')>0 OR INSTR(LOWER(CLS.BROWSER),'curl')>0 OR INSTR(LOWER(CLS.BROWSER),'wget')>0
		THEN 'other'
	ELSE 'n.a.' END AS AGENT_BOT, -- Bot
	STRFTIME('%s','now') AS UPDATE_DT
FROM CLS
LEFT OUTER JOIN DIM_USER_AGENT TRG
ON CLS.BROWSER = TRG.USER_AGENT_NK
WHERE TRG.DIM_USER_AGENT_ID IS NULL

Data Yntegraasje

Omfettet it organisearjen fan de oerdracht fan gegevens fan it bestjoeringssysteem nei it rapportaazjesysteem. Om dit te dwaan, moatte jo in poadiumtafel meitsje mei in struktuer dy't fergelykber is mei de boarne.

Ynformaasje oer websiden komt op it poadium fanút de CMS-backup yn 'e foarm fan ynfoegje oanfragen.

It laden fan de histoaryske tabel DIM_REQUEST mei basisgegevens bart yn trije stappen: nije kaaien en attributen laden, besteande bywurkje en wiske records ynsette.

Nije SCD2-opnames laden

/* Load request table SCD from master data */
INSERT INTO DIM_REQUEST (DIM_REQUEST_ID_HIST, REQUEST_NK, PAGE_TITLE, PAGE_DESCR, PAGE_KEYWORDS, DELETE_FLAG, UPDATE_DT)
WITH CLS  AS ( -- prepare keys
	SELECT
	'/' || NAME AS REQUEST_NK,
	TITLE       AS PAGE_TITLE,
	CASE WHEN DESCRIPTION = '' OR DESCRIPTION IS NULL
	     THEN 'n.a.' ELSE DESCRIPTION
	END AS PAGE_DESCR,
	CASE WHEN KEYWORDS = '' OR KEYWORDS IS NULL
	     THEN 'n.a.' ELSE KEYWORDS
	END AS PAGE_KEYWORDS
	FROM STG_CMS_MENU
	WHERE CONTENT_TYPE != 'folder' -- only web pages
	  AND PAGE_TITLE != 'n.a.' -- master data which make sense
)
/* new records from stage: CLS */
SELECT
	-1 AS DIM_REQUEST_ID_HIST,
	CLS.REQUEST_NK,
	CLS.PAGE_TITLE,
	CLS.PAGE_DESCR,
	CLS.PAGE_KEYWORDS,
	0 AS DELETE_FLAG,
	STRFTIME('%s','now') AS UPDATE_DT
FROM CLS
LEFT OUTER JOIN
 (
	SELECT
	DIM_REQUEST_ID,
	REQUEST_NK,
	PAGE_TITLE,
	PAGE_DESCR,
	PAGE_KEYWORDS
	FROM DIM_REQUEST_V_ACT
) TRG ON CLS.REQUEST_NK = TRG.REQUEST_NK
WHERE TRG.REQUEST_NK IS NULL -- no such record in data mart

SCD2-attributen bywurkje

/* Load request table SCD from master data */
INSERT INTO DIM_REQUEST (DIM_REQUEST_ID_HIST, REQUEST_NK, PAGE_TITLE, PAGE_DESCR, PAGE_KEYWORDS, DELETE_FLAG, UPDATE_DT)
WITH CLS  AS ( -- prepare keys
	SELECT
	'/' || NAME AS REQUEST_NK,
	TITLE       AS PAGE_TITLE,
	CASE WHEN DESCRIPTION = '' OR DESCRIPTION IS NULL
	     THEN 'n.a.' ELSE DESCRIPTION
	END AS PAGE_DESCR,
	CASE WHEN KEYWORDS = '' OR KEYWORDS IS NULL
	     THEN 'n.a.' ELSE KEYWORDS
	END AS PAGE_KEYWORDS
	FROM STG_CMS_MENU
	WHERE CONTENT_TYPE != 'folder' -- only web pages
	  AND PAGE_TITLE != 'n.a.' -- master data which make sense
)
/* updated records from stage: CLS and build reference to history: HIST */
SELECT
	HIST.DIM_REQUEST_ID AS DIM_REQUEST_ID_HIST,
	HIST.REQUEST_NK,
	CLS.PAGE_TITLE,
	CLS.PAGE_DESCR,
	CLS.PAGE_KEYWORDS,
	0 AS DELETE_FLAG,
	STRFTIME('%s','now') AS UPDATE_DT
FROM CLS,
     DIM_REQUEST_V_ACT TRG,
     DIM_REQUEST HIST
WHERE CLS.REQUEST_NK = TRG.REQUEST_NK
  AND TRG.DIM_REQUEST_ID = HIST.DIM_REQUEST_ID
  AND ( CLS.PAGE_TITLE != HIST.PAGE_TITLE /* changes only */
     OR CLS.PAGE_DESCR != HIST.PAGE_DESCR
     OR CLS.PAGE_KEYWORDS != HIST.PAGE_KEYWORDS )

Wiske SCD2 opnames

/* Load request table SCD from master data */
INSERT INTO DIM_REQUEST (DIM_REQUEST_ID_HIST, REQUEST_NK, PAGE_TITLE, PAGE_DESCR, PAGE_KEYWORDS, DELETE_FLAG, UPDATE_DT)
WITH CLS  AS ( -- prepare keys
	SELECT
	'/' || NAME AS REQUEST_NK,
	TITLE       AS PAGE_TITLE
	FROM STG_CMS_MENU
	WHERE CONTENT_TYPE != 'folder' -- only web pages
	  AND PAGE_TITLE != 'n.a.' -- master data which make sense
)
/*  deleted records in data mart: TRG */
SELECT
	TRG.DIM_REQUEST_ID AS DIM_REQUEST_ID_HIST,
	TRG.REQUEST_NK,
	TRG.PAGE_TITLE,
	TRG.PAGE_DESCR,
	TRG.PAGE_KEYWORDS,
	1 AS DELETE_FLAG,
	STRFTIME('%s','now') AS UPDATE_DT
FROM (
	SELECT
	DIM_REQUEST_ID,
	REQUEST_NK,
	PAGE_TITLE,
	PAGE_DESCR,
	PAGE_KEYWORDS
	FROM DIM_REQUEST_V_ACT
	WHERE PAGE_TITLE != 'n.a.' -- track master data only
	  AND DELETE_FLAG = 0 -- not already deleted
) TRG
LEFT OUTER JOIN CLS ON TRG.REQUEST_NK = CLS.REQUEST_NK
WHERE CLS.REQUEST_NK IS NULL -- no such record in stage

Elke gegevensboarne moat begelaat wurde troch in formele beskriuwing, bygelyks yn it readme.txt-bestân:

Untfanger fan de gegevens formeel / technysk: namme, e-mailadres
Data provider formeel / technysk: namme, e-mailadres
Gegevensboarne: triempaad, tsjinstnammen
Gegevens tagong ynformaasje: brûkers en wachtwurden

It gegevensstreamdiagram sil helpe by it proses fan ûnderhâld en bywurking, bygelyks yn tekstfoarm:

In triem ferpleatse. Boarne: ftp.domain.net: /logs/access.log Doel: /var/www/access.log
Lêzen op it poadium. Doel: STG_ACCESS_LOG
Laden en transformaasje. Doel: FCT_ACCESS_REQUEST_REF_HH
Laden en transformaasje. Doel: FCT_ACCESS_USER_AGENT_DD
Melde. Doel: /var/www/report.html

konklúzje

Sa beskriuwt it artikel meganismen lykas de yntegraasje fan ûnderlizzende gegevens en de ynfiering fan nije nivo's fan aggregaasje. Se binne nedich by it bouwen fan datapakhuzen om ekstra kennis te krijen en de kwaliteit fan ynformaasje te ferbetterjen.

Boarne: www.habr.com