1) Reports bound directly to tables
2) Persistent belief that the warehouse can be purchased, downloaded and installed
3) Generic db accounts with shared passwords and far reaching access
4) Tendency to pull all data into DW at the finest possible grain instead of getting clear requirements.
5) Business logic (joins, filters, definitions, security) built at report level, over and over.
6) Duplicate infrastructure. Tables, reports, BI tools, development efforts
7) Subject areas defined around business org chart, or vendor modules, not significant business nouns (eg. patient, provider, encounter, etc).
8) Persistent belief that the new version will be the last, perfect iteration.
9) Highly technical work performed by untrained or nontechnical staff
10) Persistent belief that data warehousing, BI and analytics is easy and can be done quickly by a few people.
11) Disinterested stakeholders underestimate how much work their role requires in the development process.
12) Fine grained data built into BI due to superfluous pursuit of perfection.
13) Über models designed but not implemented because legacy structures cannot be decommissioned.
14) Arguments about facts and dimensions, surrogate keys, etc. paralyze development effort and lead to inconsistent design.
15) Lack of end to end linage and impact reporting of ETL system due to metadata neglect or poor tool interoperability.
16) No data dictionary on DW. Attempts to build one bog down in ontological abstractions.
17) Database optimizer neutered because joins implemented in BI/reporting layer.
18) System complexity due to poorly integrated systems, understaffed administration, duplicate tools.
19) Customer facing report aesthetics trump hidden, complicated infrastructure needs.
20) Entire EMR history loaded every night to ensure all data correct in futile pursuit of perfection.
21) Kimball dimension "types" are unnecessary and his architecture is poorly framed and inconsistent, yet remains a dimly understood gold standard.
22) DW maintainance practice relies on drop and full refesh of tables. DW not able to reliably use surrogate keys. Source table consolidation impossible.
2) Persistent belief that the warehouse can be purchased, downloaded and installed
3) Generic db accounts with shared passwords and far reaching access
4) Tendency to pull all data into DW at the finest possible grain instead of getting clear requirements.
5) Business logic (joins, filters, definitions, security) built at report level, over and over.
6) Duplicate infrastructure. Tables, reports, BI tools, development efforts
7) Subject areas defined around business org chart, or vendor modules, not significant business nouns (eg. patient, provider, encounter, etc).
8) Persistent belief that the new version will be the last, perfect iteration.
9) Highly technical work performed by untrained or nontechnical staff
10) Persistent belief that data warehousing, BI and analytics is easy and can be done quickly by a few people.
11) Disinterested stakeholders underestimate how much work their role requires in the development process.
12) Fine grained data built into BI due to superfluous pursuit of perfection.
13) Über models designed but not implemented because legacy structures cannot be decommissioned.
14) Arguments about facts and dimensions, surrogate keys, etc. paralyze development effort and lead to inconsistent design.
15) Lack of end to end linage and impact reporting of ETL system due to metadata neglect or poor tool interoperability.
16) No data dictionary on DW. Attempts to build one bog down in ontological abstractions.
17) Database optimizer neutered because joins implemented in BI/reporting layer.
18) System complexity due to poorly integrated systems, understaffed administration, duplicate tools.
19) Customer facing report aesthetics trump hidden, complicated infrastructure needs.
20) Entire EMR history loaded every night to ensure all data correct in futile pursuit of perfection.
21) Kimball dimension "types" are unnecessary and his architecture is poorly framed and inconsistent, yet remains a dimly understood gold standard.
22) DW maintainance practice relies on drop and full refesh of tables. DW not able to reliably use surrogate keys. Source table consolidation impossible.
