How to Build a Bronze–Silver–Gold Data Pipeline in Fabric

1
00:00:00,000 --> 00:00:07,160
And here we find the enterprise data estate scattered like shy creatures hiding in separate valleys.

2
00:00:07,160 --> 00:00:14,360
Data Marts evaporating in isolation, pipelines struggling upstream, dashboards drinking from stale pools.

3
00:00:14,360 --> 00:00:23,560
Observe the fragmentation, latency, drift, duplication, then with deliberate care, a new landscape appears.

4
00:00:23,560 --> 00:00:28,520
One lake as the watershed domains as territories, workspaces as nests,

5
00:00:28,960 --> 00:00:34,440
power BI shows bright plumage, yes, but its survival depends on upstream health.

6
00:00:34,440 --> 00:00:38,360
Learn the biome, master its flows, secure its balance.

7
00:00:38,360 --> 00:00:45,720
Look closely here, map the habitats and their boundaries and the whole ecosystem begins to cooperate.

8
00:00:45,720 --> 00:00:52,760
Habitats and territories, domains, workspaces, one lake, lake house versus warehouse,

9
00:00:53,120 --> 00:01:00,880
and now observe the terrain itself. One lake rests quietly beneath everything, the unified substrate,

10
00:01:00,880 --> 00:01:10,200
a single organizational lake, a truly magnificent specimen, storing tables in delta format and files side by side,

11
00:01:10,200 --> 00:01:16,160
remove its data source and it becomes distressed, feed it correctly and the rest of the species thrive.

12
00:01:16,160 --> 00:01:21,640
One lake doesn't demand one giant clearing, it governs the water table, territories still matter.

13
00:01:22,280 --> 00:01:28,440
Domains define responsibility boundaries, sales, finance, operations, HR marketing.

14
00:01:28,440 --> 00:01:31,880
Each domain curates its own habitat and rituals.

15
00:01:31,880 --> 00:01:35,320
Upset this balance and chaos spreads swiftly.

16
00:01:35,320 --> 00:01:41,440
With domains ownership is legible, lineage is easier to track, stewardship has a place to live.

17
00:01:41,440 --> 00:01:48,600
Domains reduce sprawl not by fences alone, but by shared expectations of care, data contracts,

18
00:01:48,680 --> 00:01:52,160
refresh discipline, label usage, incident paths.

19
00:01:52,160 --> 00:02:00,280
Within each domain, workspaces form the practical nests they tie to capacity, the energy budget of the habitat.

20
00:02:00,280 --> 00:02:03,800
Dev workspaces are experimental clearings where make us roam.

21
00:02:03,800 --> 00:02:08,160
Prod workspaces are protected breeding grounds for hardened species.

22
00:02:08,160 --> 00:02:14,000
The bronze, silver, gold, layering appears here like soil strata.

23
00:02:14,960 --> 00:02:19,400
Bronze captures raw sediment exactly as it lands.

24
00:02:19,400 --> 00:02:22,680
Silver standardises and removes debris.

25
00:02:22,680 --> 00:02:27,040
Gold is curated forage, ready for broad consumption.

26
00:02:27,040 --> 00:02:31,680
When a gold patch is trampled with ad hoc changes, the herd becomes nervous.

27
00:02:31,680 --> 00:02:37,920
Keep trails clear, promotion pipelines, approvals and versioning restore calm.

28
00:02:37,920 --> 00:02:42,800
Now observe the two complementary shelters, Lake House and Warehouse.

29
00:02:43,000 --> 00:02:48,520
The Lake House is open terrain, files, delta tables and shortcuts to distant groves,

30
00:02:48,520 --> 00:02:58,080
analysts, engineers and data scientists wander together here, leaving tracks in power query, spark or notebooks.

31
00:02:58,080 --> 00:03:02,960
Shortcuts act as territory bridges, they carry responsibility with them.

32
00:03:02,960 --> 00:03:10,240
If a sales lake house shortcuts to a finance gold table, finances governance flows across that bridge.

33
00:03:10,680 --> 00:03:13,720
No duplication, no secret feedings at night.

34
00:03:13,720 --> 00:03:21,120
The Warehouse is more structured shelter, a sequel first habitat where relational creatures prefer orderly paths.

35
00:03:21,120 --> 00:03:25,520
It provides a sturdy canopy for star schema dwellers and familiar tooling.

36
00:03:25,520 --> 00:03:30,000
T.S. school, views, stored procedures and performance dashboards.

37
00:03:30,000 --> 00:03:33,360
The Warehouse complements the Lake House, not replaces it.

38
00:03:33,360 --> 00:03:37,920
Many flocks move between them, ingest and refine in the Lake House,

39
00:03:38,240 --> 00:03:46,320
then expose steady, sequel endpoints via the warehouse when the herd needs consistent lanes and indexing behavior.

40
00:03:46,320 --> 00:03:52,120
Above both shelters, the semantic model becomes the shared language of the ecosystem.

41
00:03:52,120 --> 00:03:58,320
Measures, relationships and roles describe house species relate regardless of where they forage.

42
00:03:58,320 --> 00:04:01,440
Created ones, reused often.

43
00:04:01,440 --> 00:04:06,280
This shared vocabulary stops the chatter of conflicting definitions.

44
00:04:06,600 --> 00:04:09,400
Revenue means the same sound in every clearing.

45
00:04:09,400 --> 00:04:13,080
When aligned to one lake, these models can speak in direct lake,

46
00:04:13,080 --> 00:04:19,920
grazing directly on delta tables without import rituals, reports and dashboards, pre- and display,

47
00:04:19,920 --> 00:04:25,280
but their vitality depends on that semantic tongue and the health of upstream flows.

48
00:04:25,280 --> 00:04:31,480
Capacity meanwhile is the climate, too weak, and storms of refreshed backlogs roll in.

49
00:04:31,480 --> 00:04:36,400
Too strong without governance and waste lingers like standing water.

50
00:04:36,880 --> 00:04:39,400
Assign capacities to domains thoughtfully.

51
00:04:39,400 --> 00:04:47,040
Separate dev from prod to shield the nesting grounds, monitor lineage and usage telemetry as migration patterns,

52
00:04:47,040 --> 00:04:51,040
who drinks where, how often, and at what cost.

53
00:04:51,040 --> 00:04:59,800
Then, with remarkable precision, adjust quotas, schedules and caching to keep the herd balanced.

54
00:05:00,000 --> 00:05:06,480
Finally, watch how purview sensitivity labels and one lake security mark the food chain.

55
00:05:06,480 --> 00:05:11,520
Labels travel with the data, feather tags that persist across tools.

56
00:05:11,520 --> 00:05:17,640
One lake item level and column level controls act like gentle wardens at the edge of the glade,

57
00:05:17,640 --> 00:05:21,240
allowing only the right eyes to see the right leaves.

58
00:05:21,240 --> 00:05:25,400
Role level security narrows the view further.

59
00:05:25,840 --> 00:05:30,120
A regional manager sees only her watershed, never the whole river.

60
00:05:30,120 --> 00:05:32,360
This is not restriction for its own sake.

61
00:05:32,360 --> 00:05:33,960
It is habitat stability.

62
00:05:33,960 --> 00:05:35,560
It preserves trust.

63
00:05:35,560 --> 00:05:37,320
So the map is clear.

64
00:05:37,320 --> 00:05:43,240
Domains as territories, workspaces as nests bound to capacity,

65
00:05:43,240 --> 00:05:48,640
bronze silver, gold as soil, lake house as open range,

66
00:05:48,640 --> 00:05:55,520
warehouse as structured shelter and the semantic model as the language that binds the species.

67
00:05:55,960 --> 00:06:00,880
Keep the bridges honest, shortcuts must carry stewardship, keep the climate steady.

68
00:06:00,880 --> 00:06:06,760
Capacities must fit the season, maintain the tags, labels and rolls must follow the herd.

69
00:06:06,760 --> 00:06:11,760
Once this balance holds, power beies, bright displays are not a miracle,

70
00:06:11,760 --> 00:06:15,240
just the natural outcome of a healthy ecosystem.

71
00:06:15,240 --> 00:06:21,880
Rivers, streams and currents, pipelines, data flows, gen 2 ingestion governance,

72
00:06:21,880 --> 00:06:23,840
now observe the water.

73
00:06:24,280 --> 00:06:26,720
Every habitat survives on its rivers,

74
00:06:26,720 --> 00:06:29,600
calm tributaries from sass headwaters,

75
00:06:29,600 --> 00:06:35,080
snow melt from CRM plateaus, ground water seeping from files and databases.

76
00:06:35,080 --> 00:06:39,200
State effectory provides the channels over 300 connectors,

77
00:06:39,200 --> 00:06:43,800
each a reliable spring, bring flow down to one lake shore.

78
00:06:43,800 --> 00:06:50,360
But tread carefully here, a river without banks becomes a floodplain, set the banks first.

79
00:06:50,560 --> 00:06:55,240
Data flows gen 2 appears as a gentle well-worn stream for analysts.

80
00:06:55,240 --> 00:07:01,680
It uses the familiar power query engine, shaping stones into smooth forms as it travels,

81
00:07:01,680 --> 00:07:07,840
dropper CSV into files and with a few transformations, data types,

82
00:07:07,840 --> 00:07:13,240
merges, conditional branches, the stream settles into delta tables in the lake house.

83
00:07:13,240 --> 00:07:20,040
No heavy machines required, the herd drinks sooner, shortcuts and mirroring behave like engineered canals.

84
00:07:20,280 --> 00:07:23,760
A shortcut carries water from a distant lake without scooping buckets,

85
00:07:23,760 --> 00:07:26,480
zero copy, lineage intact.

86
00:07:26,480 --> 00:07:32,280
Mirroring, meanwhile, runs a near real-time flume from operational databases,

87
00:07:32,280 --> 00:07:37,000
sickle, post-gress, cosmos, into curated delta beds.

88
00:07:37,000 --> 00:07:42,240
The upstream remains undisturbed, the downstream receives fresh flow.

89
00:07:42,240 --> 00:07:47,040
Once the stream hits one lake, every species can sip, spark,

90
00:07:47,200 --> 00:07:51,680
sickle, power BI, all without building separate ditches.

91
00:07:51,680 --> 00:07:55,920
The bronze, silver, gold gradient is hydraulic logic.

92
00:07:55,920 --> 00:08:01,200
Bronze is raw capture, silt, sticks, the full sediment of reality.

93
00:08:01,200 --> 00:08:06,760
Silver settles the particulates, conformance, standardized types, keys aligned.

94
00:08:06,760 --> 00:08:13,760
Gold is clear, curated water, business ready, predictable depth, safe crossing points.

95
00:08:13,760 --> 00:08:16,840
Resist the urge to shortcut the settling ponds.

96
00:08:17,320 --> 00:08:22,440
When debris isn't given space to sink, it drifts into displays and cuts the hooves.

97
00:08:22,440 --> 00:08:24,360
Telemetry is the river ranger.

98
00:08:24,360 --> 00:08:28,400
Lineage shows the map, which inlet feeds, which pool.

99
00:08:28,400 --> 00:08:33,920
Catalog marks the signposts, discoverable names, descriptions, ownership,

100
00:08:33,920 --> 00:08:38,400
drift detection listens for a new sound in the current, a column missing,

101
00:08:38,400 --> 00:08:40,480
a type changed, a widened pipe.

102
00:08:40,480 --> 00:08:44,760
When you hear that change, adjust the weirs, update the schema, alert the stewards,

103
00:08:44,760 --> 00:08:45,920
revise the schedule.

104
00:08:46,040 --> 00:08:47,160
Remember this number.

105
00:08:47,160 --> 00:08:53,000
One owner per flow, shared streams without a name steward dry out or burst banks,

106
00:08:53,000 --> 00:08:56,840
back pressure matters, not every river should flow forever.

107
00:08:56,840 --> 00:09:02,080
Set retention at each stage, bronze for weeks to replay incidents,

108
00:09:02,080 --> 00:09:07,880
silver for months to ensure stability, gold for as long as governance and cost degree.

109
00:09:07,880 --> 00:09:10,640
Quotas prevent overgrazing.

110
00:09:10,640 --> 00:09:14,400
Concurrency caps keep turbines from overheating.

111
00:09:15,040 --> 00:09:20,920
Schedule cadence reflects the herd's thirst, hourly for inventory, daily for sales,

112
00:09:20,920 --> 00:09:25,600
real time for fraud, everything changes when a stream is matched to its purpose.

113
00:09:25,600 --> 00:09:30,280
Now observe schedule ownership, a flow without a caretaker slips time.

114
00:09:30,280 --> 00:09:35,160
Assign maintainers at the domain level, the sales steward owns sales flows,

115
00:09:35,160 --> 00:09:36,960
publish contact points.

116
00:09:36,960 --> 00:09:42,600
When a flood arrives or a drought is sensed, responders know which banks to raise.

117
00:09:42,720 --> 00:09:48,280
Use deployment pipelines as migration ladders, dev to test to prod with approvals,

118
00:09:48,280 --> 00:09:51,680
parameterized connections and version definitions.

119
00:09:51,680 --> 00:09:58,560
The fish climb reliably, the water remains clear below, a micro story from the field path.

120
00:09:58,560 --> 00:10:02,000
An analyst landing customer files uses data flows,

121
00:10:02,000 --> 00:10:05,360
Gen 2 to promote them from bronze to silver.

122
00:10:05,360 --> 00:10:09,480
Hedda fixes age calculation, segmentation banding.

123
00:10:09,760 --> 00:10:15,800
The stream writes to delta in the lake house and a semantic model downstream switches its intake

124
00:10:15,800 --> 00:10:20,080
to the new silver pool, reports bright and without a full reservoir swap.

125
00:10:20,080 --> 00:10:22,240
That's the power of aligned channels.

126
00:10:22,240 --> 00:10:27,120
Governance watches the spillways, define who can create shortcuts,

127
00:10:27,120 --> 00:10:30,400
bridges must carry the originating domains rules.

128
00:10:30,400 --> 00:10:34,360
Register every river in the catalog with sensitivity labels.

129
00:10:34,360 --> 00:10:37,480
Public internal confidential, highly confidential,

130
00:10:37,680 --> 00:10:40,640
so the tags travel downstream into power BI and office.

131
00:10:40,640 --> 00:10:43,880
One lake security controls outbound tributaries.

132
00:10:43,880 --> 00:10:47,680
Spark cannot quietly carve a tunnel to an outside valley.

133
00:10:47,680 --> 00:10:51,040
Upset this balance and chaos spreads swiftly.

134
00:10:51,040 --> 00:10:53,640
Finally, cost is the tide, monitor,

135
00:10:53,640 --> 00:10:56,200
agress compute minutes, failed refresh retreats.

136
00:10:56,200 --> 00:10:59,760
If a nightly deluge moves, one pebble reduce the flow.

137
00:10:59,760 --> 00:11:04,400
If a trickle starves the meadow, widen the gate with remarkable precision,

138
00:11:04,480 --> 00:11:09,960
tune pipelines, data flows and mirroring frequency until the ecosystem hums.

139
00:11:09,960 --> 00:11:13,720
Not silent, not roaring, simply steady.

140
00:11:13,720 --> 00:11:17,600
Then the bright feathered species will return each morning,

141
00:11:17,600 --> 00:11:20,320
drink and display without panic.

142
00:11:20,320 --> 00:11:26,360
The power BI species semantic models direct lake reports dashboards

143
00:11:26,360 --> 00:11:30,680
now observe the bright feathered species as it steps to the water's edge.

144
00:11:30,880 --> 00:11:34,960
Power BI is not the lake. It is the display bird that survives

145
00:11:34,960 --> 00:11:37,480
because the current's upstream are clean.

146
00:11:37,480 --> 00:11:40,360
Its coal carries far across the valley,

147
00:11:40,360 --> 00:11:43,720
but only when the shared language beneath it is sound.

148
00:11:43,720 --> 00:11:48,400
That language is the semantic model measures relationships and roles.

149
00:11:48,400 --> 00:11:52,560
These are its grammar and tone, define the grain of each table with care.

150
00:11:52,560 --> 00:11:58,920
A fact table that forgets its keys becomes a confused creature darting in circles.

151
00:11:59,400 --> 00:12:04,880
A dimension that repeats names without surrogate keys attracts predators called ambiguity.

152
00:12:04,880 --> 00:12:08,320
When the model is tuned, star-shaped clear,

153
00:12:08,320 --> 00:12:11,080
relationships directional and intentional,

154
00:12:11,080 --> 00:12:15,440
measures composed from base columns rather than layered on other measures,

155
00:12:15,440 --> 00:12:18,320
the species sings with confidence.

156
00:12:18,320 --> 00:12:21,640
Revenue, margin, win rate,

157
00:12:21,640 --> 00:12:26,360
each coal precise, repeatable and understood across clearings,

158
00:12:26,760 --> 00:12:31,240
then with remarkable precision watch direct lake behavior.

159
00:12:31,240 --> 00:12:35,080
Where once the species require daily feeding rituals,

160
00:12:35,080 --> 00:12:38,080
imports a dawn, refresh cues at dusk,

161
00:12:38,080 --> 00:12:42,400
direct lake frees it to graze directly on delta tables in one lake.

162
00:12:42,400 --> 00:12:45,320
No sacks of cached grain to lug around,

163
00:12:45,320 --> 00:12:48,760
no stale fodder lingering after the weather changes.

164
00:12:48,760 --> 00:12:51,360
The preconditions matter, delta format,

165
00:12:51,360 --> 00:12:54,960
compatible model structure and access to the right pools.

166
00:12:55,320 --> 00:12:59,520
Satisfy these and visuals respond with the snap of import

167
00:12:59,520 --> 00:13:01,800
and the freshness of live water.

168
00:13:01,800 --> 00:13:06,360
The grazing path shortens the energy returns to the display.

169
00:13:06,360 --> 00:13:10,760
Now a brief demonstration ritual, demo A in the field notes,

170
00:13:10,760 --> 00:13:13,680
the keeper creates a lake house in the sales domain,

171
00:13:13,680 --> 00:13:18,640
a shortcut bridges to curated internet sales in silver with dimensions,

172
00:13:18,640 --> 00:13:24,320
date, geography, product, customer crossing from endorsed groves,

173
00:13:24,720 --> 00:13:29,360
a default semantic model appears the keeper opens it sketches relationships,

174
00:13:29,360 --> 00:13:34,400
many to one from internet sales to date, customer, geography, product,

175
00:13:34,400 --> 00:13:37,880
one base measure rises total sales,

176
00:13:37,880 --> 00:13:43,080
exos, some sales amount formats it as currency,

177
00:13:43,080 --> 00:13:49,720
a small grooming act that matters, then power bi desktop connects using direct lake,

178
00:13:49,720 --> 00:13:54,680
the species lifts a map visuals by country with total

179
00:13:54,680 --> 00:13:59,000
sales, a bar of age breakdown from the new silver customer table,

180
00:13:59,000 --> 00:14:03,320
a trend line by month, no refresh ritual, no import feed,

181
00:14:03,320 --> 00:14:06,680
the bird eats from the stream and displays at once.

182
00:14:06,680 --> 00:14:11,480
But tread carefully here visibility invites predators called misconception.

183
00:14:11,480 --> 00:14:13,600
Three approach the clearing often.

184
00:14:13,600 --> 00:14:16,360
First, fabric replaces power bi.

185
00:14:16,360 --> 00:14:20,800
Not so, power bi remains the species that displays the truth.

186
00:14:21,240 --> 00:14:26,920
Fabric is the ecosystem that sustains it. Second, one lake means one giant workspace.

187
00:14:26,920 --> 00:14:33,560
No, one lake is a watershed, not a single nest domains and workspaces still govern territory.

188
00:14:33,560 --> 00:14:37,240
Third, fabric is for engineers, not bi people.

189
00:14:37,240 --> 00:14:40,120
Observe data flows gen two and shortcuts.

190
00:14:40,120 --> 00:14:42,560
They make engineering approachable.

191
00:14:42,560 --> 00:14:44,880
The stepping stones are flat.

192
00:14:44,880 --> 00:14:51,080
The currents slow enough for analysts to cross reports and dashboards are the courtship dance.

193
00:14:51,520 --> 00:14:58,240
Reports allow exploration, slices as branches, drill through paths as burrow trails,

194
00:14:58,240 --> 00:15:04,360
bookmarks as ritual postures, dashboards, pin the sightings for daily patrols,

195
00:15:04,360 --> 00:15:07,400
a quick scan of the meadow at sunrise.

196
00:15:07,400 --> 00:15:11,000
Love from the business depends on upstream honesty.

197
00:15:11,000 --> 00:15:15,200
When a manager slices by region and the totals ripple oddly,

198
00:15:15,200 --> 00:15:17,360
trust scatters like starlings.

199
00:15:18,040 --> 00:15:23,120
The remedy isn't louder feathers, it's a cleaner intake, fix the silver conformance,

200
00:15:23,120 --> 00:15:29,840
refine the role security, clarify the measure, roles, observe their quiet influence.

201
00:15:29,840 --> 00:15:34,000
Role level security narrows the species field of view naturally.

202
00:15:34,000 --> 00:15:39,440
A regional lead sees only her watershed, a product director sees only his herd.

203
00:15:39,440 --> 00:15:43,440
Role logic belongs in the semantic model close to the language.

204
00:15:43,440 --> 00:15:47,000
So every report using that tongue inherits the same sight lines.

205
00:15:47,560 --> 00:15:50,000
Do not laminate security in visuals.

206
00:15:50,000 --> 00:15:52,640
That is camouflage, not protection.

207
00:15:52,640 --> 00:15:58,600
Everything changes when the semantic model becomes a shared species, not a private one.

208
00:15:58,600 --> 00:16:02,280
Publish it as a certified model in the domain's gold workspace.

209
00:16:02,280 --> 00:16:08,920
Invite reuse, the herd stops reinventing revenue and starts debating outliers instead.

210
00:16:08,920 --> 00:16:16,360
Performance follows a single well tuned model scales better than many hidden,

211
00:16:16,560 --> 00:16:19,440
slightly different cousins nibbling at the same patch.

212
00:16:19,440 --> 00:16:21,360
Finally, watch the energy budget.

213
00:16:21,360 --> 00:16:26,600
Even direct lake can tire if star schemas are muddled or cardinality explodes.

214
00:16:26,600 --> 00:16:31,080
Prune wide-text columns from facts aggregate where business questions allow,

215
00:16:31,080 --> 00:16:35,400
use composite models only when the migration route demands an interim roost.

216
00:16:35,400 --> 00:16:42,360
Lineage will show who drinks where, use it to rescue overgraced clearings and to celebrate efficient trails.

217
00:16:42,360 --> 00:16:44,560
So the lesson is simple, handled with care.

218
00:16:44,920 --> 00:16:46,720
Nurture the semantic language.

219
00:16:46,720 --> 00:16:49,680
Let direct lake shorten the path to fresh water.

220
00:16:49,680 --> 00:16:54,760
Stage the dances, reports and dashboards only after the channels run clear.

221
00:16:54,760 --> 00:16:58,000
And the bright feathered species will not just survive here.

222
00:16:58,000 --> 00:17:02,040
It will become the signal that the entire ecosystem is healthy.

223
00:17:02,040 --> 00:17:06,600
Predators in protection, security, compliance, life cycle.

224
00:17:06,600 --> 00:17:09,600
Now observe what prowls at the edge of the clearing.

225
00:17:09,600 --> 00:17:14,480
Visibility draws attention and attention invites predators.

226
00:17:14,840 --> 00:17:16,960
Curiosity from the herd is welcome.

227
00:17:16,960 --> 00:17:21,720
Exfiltration, tampering, and accidental oversharing are not.

228
00:17:21,720 --> 00:17:26,920
A healthy ecosystem survives because its protections are woven into the habitat,

229
00:17:26,920 --> 00:17:29,640
not draped over it after the tracks are set.

230
00:17:29,640 --> 00:17:31,320
Territory rights come first.

231
00:17:31,320 --> 00:17:33,720
Workspace roles are the boundary stones.

232
00:17:33,720 --> 00:17:36,920
Atman, member, contributor, viewer.

233
00:17:36,920 --> 00:17:38,520
Assign them with intention.

234
00:17:38,520 --> 00:17:40,800
Dev clearings allow more footprints.

235
00:17:40,800 --> 00:17:42,960
Prod remains a protected breeding ground.

236
00:17:43,560 --> 00:17:47,240
Upset this balance and chaos spread swiftly.

237
00:17:47,240 --> 00:17:51,560
Add hoc edits in gold unexplained metric shifts,

238
00:17:51,560 --> 00:17:53,400
nocturnal refresh storms.

239
00:17:53,400 --> 00:17:55,960
Restore calm with deployment pipelines.

240
00:17:55,960 --> 00:17:58,320
Dev test, prod.

241
00:17:58,320 --> 00:18:02,000
So species migrate predictably, not in panicked flocks.

242
00:18:02,000 --> 00:18:06,240
Row level security narrows the field of view to the proper watershed.

243
00:18:06,240 --> 00:18:09,720
Define it in the semantic model, the language layer.

244
00:18:09,720 --> 00:18:13,160
So every report inherits the same sidelines.

245
00:18:13,240 --> 00:18:16,120
Roads travel with the model like a quiet scent.

246
00:18:16,120 --> 00:18:22,480
Visuals need no camouflage when geography, business unit or customer segment governs access.

247
00:18:22,480 --> 00:18:26,520
Encode those rules as filters tied to identity.

248
00:18:26,520 --> 00:18:28,960
The bird sees only its valley.

249
00:18:28,960 --> 00:18:31,400
The valley remains undisturbed.

250
00:18:31,400 --> 00:18:37,240
Object and column level control step in when finer leaves must remain hidden.

251
00:18:37,840 --> 00:18:46,760
Sensitive columns, national IDs, credit card tokens, salary bands can be masked or excluded per roll.

252
00:18:46,760 --> 00:18:51,280
In the warehouse shelter, SQL permissions and views provide orderly lanes.

253
00:18:51,280 --> 00:18:58,760
In the lake house range, one lakes item, folder and column level security act like gentle wardens at the trailhead,

254
00:18:58,760 --> 00:19:03,400
granting passage to the right foragers while keeping seedlings from curious mouths.

255
00:19:03,400 --> 00:19:05,520
Labels mark the food chain.

256
00:19:05,880 --> 00:19:11,360
Per view sensitivity labels, public internal confidential, highly confidential,

257
00:19:11,360 --> 00:19:19,560
are feather tags that travel with the data across tools, power, BI, Excel, teams, even out to office co-authoring.

258
00:19:19,560 --> 00:19:24,840
Apply them at the workspace where possible, let inheritance simplify care.

259
00:19:24,840 --> 00:19:34,920
When the label changes, downstream creatures adjust posture, export restrictions, water marking, data loss prevention, without fresh instruction calls.

260
00:19:35,320 --> 00:19:40,400
This is compliance behaving like nature, always on, seldom theatrical.

261
00:19:40,400 --> 00:19:48,360
Observe one lake protections guarding the waters edge, outbound access controls prevent spark from carving secret channels to foreign valleys.

262
00:19:48,360 --> 00:19:52,640
Private link keeps traffic on trusted paths away from open plains.

263
00:19:52,640 --> 00:19:56,760
Customer managed keys were required at another carapace to the shell.

264
00:19:56,760 --> 00:19:58,280
None of this is ornamental.

265
00:19:58,280 --> 00:20:01,320
It is quiet strength, invisible until disturbed.

266
00:20:01,320 --> 00:20:05,000
Audit, lineage and usage are the ecosystems hearing.

267
00:20:05,480 --> 00:20:10,880
Linear shows who drank where, upstream to downstream, table to model to report.

268
00:20:10,880 --> 00:20:17,160
When numbers shift at dawn, follow the tracks upstream, usage metrics reveal grazing pressure,

269
00:20:17,160 --> 00:20:23,120
which reports the herd visits, which models carry the weight, which queries stir the silt.

270
00:20:23,120 --> 00:20:32,240
Anomalous patterns, sudden exports, unusual volumes, after hours, flurries, raise a soft alarm.

271
00:20:32,800 --> 00:20:37,200
The stewards approach, not with panic, but with practiced curiosity.

272
00:20:37,200 --> 00:20:41,160
Now, a brief demonstration ritual, demo B in the field notes.

273
00:20:41,160 --> 00:20:51,920
In the sales gold workspace, the keeper opens the semantic model and assigns a region manager role, region, user principle name.

274
00:20:51,920 --> 00:20:53,880
MAP through a lookup.

275
00:20:53,880 --> 00:21:00,800
Next, the sensitivity label, confidential, is applied at the workspace, feathering every item within.

276
00:21:01,160 --> 00:21:06,720
Back in power BI, two viewers step to the bank, one from Emia, one from North America.

277
00:21:06,720 --> 00:21:09,400
The same report shows different valleys.

278
00:21:09,400 --> 00:21:15,960
The watermark appears, export is limited, no brittle visual level filters, no duplicated reports by region.

279
00:21:15,960 --> 00:21:19,000
The protection resides in the habitat itself.

280
00:21:19,000 --> 00:21:21,760
Life cycle completes the defense.

281
00:21:21,760 --> 00:21:30,360
Source control ties definitions to a living chronicle, who changed the measure when a relationship pivoted, why the refresh cadence shifted.

282
00:21:30,880 --> 00:21:36,120
Deployment pipelines carry artifacts up the migration ladders with parameters and checks.

283
00:21:36,120 --> 00:21:42,920
Dev endpoints here, prod endpoints there, ensuring the fish released upstream are the same ones expected downstream.

284
00:21:42,920 --> 00:21:45,480
Rollback is a tide, not a landslide.

285
00:21:45,480 --> 00:21:48,720
Governance is stewardship, not spectacle.

286
00:21:48,720 --> 00:21:54,440
Name the stewards in each domain, humans with ears to the river, publish contact points.

287
00:21:54,440 --> 00:21:56,840
Define incident paths.

288
00:21:57,360 --> 00:22:03,280
Schema drift at bronze, conformance, breach at silver, semantic change at gold.

289
00:22:03,280 --> 00:22:06,440
Schedule ownership belongs to the domain that drinks the water.

290
00:22:06,440 --> 00:22:13,640
Back pressure rules, retention horizons, capacity quotas, all documented, all visible in the catalog.

291
00:22:13,640 --> 00:22:17,000
Finally, remember this quiet truth.

292
00:22:17,000 --> 00:22:20,160
Protection does not slow the ecosystem.

293
00:22:20,160 --> 00:22:22,880
It grants it the confidence to move.

294
00:22:23,440 --> 00:22:29,200
When roles live in the language labels travel with the grain and migrations climb predictable ladders,

295
00:22:29,200 --> 00:22:32,440
the bright feathered species can display without fear.

296
00:22:32,440 --> 00:22:40,880
The predators stay at the tree line, wary of a habitat that knows itself and defends itself, with grace.

297
00:22:40,880 --> 00:22:46,120
Symbiotic species, copilot and AI, assistance.

298
00:22:46,120 --> 00:22:49,520
Now, observe the helper at the edge of the glade.

299
00:22:49,880 --> 00:22:57,560
Copilot arrives as a symbiant, not a predator, not a replacement, a diligent cleaner that lives on the same branch

300
00:22:57,560 --> 00:23:01,000
as the bright feathered species grooming what it cannot reach.

301
00:23:01,000 --> 00:23:06,120
It learns the rhythms of the habitat and whispers useful suggestions,

302
00:23:06,120 --> 00:23:09,960
but only when the water is clear and the language is consistent.

303
00:23:09,960 --> 00:23:11,520
Look closely here.

304
00:23:11,520 --> 00:23:15,360
In the Lake House, copilot drafts ingestion steps,

305
00:23:15,640 --> 00:23:21,080
proposes power query transforms, suggests column types, even scaffolds,

306
00:23:21,080 --> 00:23:24,120
a data flows gen 2 stream from bronze to silver.

307
00:23:24,120 --> 00:23:28,120
It reads the strata and proposes orderly layers.

308
00:23:28,120 --> 00:23:37,200
In the semantic model, it nudges measure definitions, repairs relationships and translates a business question into a DAC sketch

309
00:23:37,200 --> 00:23:41,240
ready for the keeper's touch, then with remarkable precision.

310
00:23:41,760 --> 00:23:47,520
It cites lineage so the steward can verify provenance before anything migrates upstream,

311
00:23:47,520 --> 00:23:49,040
but tread carefully here.

312
00:23:49,040 --> 00:23:52,040
Assymbiant thrives in a clean habitat.

313
00:23:52,040 --> 00:23:58,640
In disordered terrain, its guidance becomes uncertain, like a bird calling from fog.

314
00:23:58,640 --> 00:24:04,000
When tables lack keys, when labels are missing, when domains blur responsibility,

315
00:24:04,000 --> 00:24:11,640
copilot's confidence wanes, restore clarity, schemas named, ownership listed, sensitivity labels applied,

316
00:24:11,640 --> 00:24:13,920
and the symbiote grows precise again.

317
00:24:13,920 --> 00:24:21,480
Governance aligned prompts strengthen the bond, use endorsed silver sources, respect confidential tags,

318
00:24:21,480 --> 00:24:26,320
limit output to fields approved for export, observe permission awareness.

319
00:24:26,320 --> 00:24:29,280
Copilot only sees what the identity can see.

320
00:24:29,280 --> 00:24:34,840
Row level security closes its view naturally, item level controls bound its reach.

321
00:24:34,840 --> 00:24:36,960
The result is a polite companion.

322
00:24:36,960 --> 00:24:40,000
It works within the valley, never over the ridge.

323
00:24:40,400 --> 00:24:42,680
Agents extend the behavior further.

324
00:24:42,680 --> 00:24:48,800
Small, specialized symbiants that can traverse mirrored sources align facts across domains

325
00:24:48,800 --> 00:24:53,120
and return with stitched insights, still wearing the right labels.

326
00:24:53,120 --> 00:24:59,880
A micro story from the clearing, a maker asks, show me win rate by segment and region.

327
00:24:59,880 --> 00:25:08,520
Copilot scans the certified model, recognizes the base measures and composes a visual layout with proper filters.

328
00:25:08,560 --> 00:25:13,240
It highlights that a geography lookup is missing for a new emea subregion

329
00:25:13,240 --> 00:25:16,840
and offers the data flows, gen 2 step to correct it.

330
00:25:16,840 --> 00:25:22,840
The keeper approves, the stream is published to silver, and the semantic measure continues unchanged.

331
00:25:22,840 --> 00:25:25,480
No drama, just helpful grooming.

332
00:25:25,480 --> 00:25:27,600
Finally, health checks.

333
00:25:27,600 --> 00:25:34,320
Copilot audit schedules, flags, long refresh cues, and points at gold tables that seldom receive visitors.

334
00:25:34,320 --> 00:25:36,600
Consider archiving, it suggests.

335
00:25:36,760 --> 00:25:38,760
Not an order, a gentle nudge.

336
00:25:38,760 --> 00:25:42,440
In this ecosystem, intelligence isn't a showy roar.

337
00:25:42,440 --> 00:25:50,120
It is quiet, context-aware assistance that keeps the trails neat so the species can display with confidence.

338
00:25:50,120 --> 00:25:55,200
Field path, sales analytics journey plus 30-day governance starter,

339
00:25:55,200 --> 00:25:57,280
and here we follow a migration.

340
00:25:57,280 --> 00:25:58,840
Any flock can trace.

341
00:25:58,840 --> 00:26:01,920
The sales domain prepares its path.

342
00:26:02,520 --> 00:26:10,840
CRM headwaters, lake house, data flows, gen 2, semantic model, power BI display.

343
00:26:10,840 --> 00:26:16,120
A calm, repeatable journey, watched by stewards and guarded by labels.

344
00:26:16,120 --> 00:26:17,840
First, ingestion.

345
00:26:17,840 --> 00:26:25,160
A data factory connector draws daily changes from CRM, accounts, opportunities, activities,

346
00:26:25,160 --> 00:26:27,840
into the sales lake house as bronze delta.

347
00:26:28,160 --> 00:26:36,240
Lineage records the inlet, sensitivity is marked internal, and ownership is assigned to the sales steward.

348
00:26:36,240 --> 00:26:39,680
Backpressure rules retain 14 days in bronze for replay.

349
00:26:39,680 --> 00:26:41,520
Second, refinement.

350
00:26:41,520 --> 00:26:42,680
Data flows.

351
00:26:42,680 --> 00:26:44,400
Gen 2 promotes to silver.

352
00:26:44,400 --> 00:26:47,360
Data types fixed, surrogate keys created.

353
00:26:47,360 --> 00:26:50,400
Closed one logic standardized, a date table conformed.

354
00:26:50,400 --> 00:26:53,560
The customer's stream merges marketing segments.

355
00:26:53,560 --> 00:26:56,480
The activity stream derives touch frequency.

356
00:26:56,960 --> 00:27:02,880
Each flow has one named maintainer, a schedule posted, and drift detection configured.

357
00:27:02,880 --> 00:27:05,480
The silver watershed becomes predictable.

358
00:27:05,480 --> 00:27:07,880
Third, language.

359
00:27:07,880 --> 00:27:14,440
A sales semantic model in the gold workspace defines grain and relationships.

360
00:27:14,440 --> 00:27:18,640
Internet sales fact-to-date customer product geography.

361
00:27:18,640 --> 00:27:21,280
Measures sing clearly.

362
00:27:21,280 --> 00:27:26,320
Total sales, pipeline, win rate, average deal size.

363
00:27:26,480 --> 00:27:28,160
Quotar attainment.

364
00:27:28,160 --> 00:27:31,960
Row-level security narrow sidelines by region.

365
00:27:31,960 --> 00:27:34,960
A product role offers a different view.

366
00:27:34,960 --> 00:27:37,120
The model is certified for reuse.

367
00:27:37,120 --> 00:27:43,520
Per view applies confidential at the workspace, so the tag travels into power, BI, and office.

368
00:27:43,520 --> 00:27:45,520
Fourth, display.

369
00:27:45,520 --> 00:27:47,840
Reports show the meadow at sunrise.

370
00:27:47,840 --> 00:27:55,120
A pipeline funnel by stage, a trend by month, a map by region, a table of at-risk deals by inactivity.

371
00:27:55,280 --> 00:27:57,560
Direct Lake shortens the path.

372
00:27:57,560 --> 00:28:00,800
Visuals graze on silver and gold delta.

373
00:28:00,800 --> 00:28:03,400
Managers filter confidently.

374
00:28:03,400 --> 00:28:05,520
Numbers hold their shape.

375
00:28:05,520 --> 00:28:08,720
Now the 30-day baseline settles the habitat.

376
00:28:08,720 --> 00:28:15,520
Day 1-7, established domains, sales, finance, operations, HR, marketing,

377
00:28:15,520 --> 00:28:19,760
and a point data stewards with published contact paths.

378
00:28:19,760 --> 00:28:23,320
Create dev and prod workspaces per domain,

379
00:28:23,480 --> 00:28:26,840
block personal fabric workspaces for workloads.

380
00:28:26,840 --> 00:28:31,680
Day 8-14, adopt naming that reads like signposts.

381
00:28:31,680 --> 00:28:36,760
Sales, bronze, lake house, sales, gold, semantic model.

382
00:28:36,760 --> 00:28:41,160
Register streams in the catalog with owners, descriptions and labels.

383
00:28:41,160 --> 00:28:47,880
Day 15-21, implement bronze silver gold rituals with deployment pipelines.

384
00:28:47,880 --> 00:28:53,400
Dev, test, prod, parameterized connections, approvals,

385
00:28:53,480 --> 00:28:55,640
and versioned definitions.

386
00:28:55,640 --> 00:29:00,040
Define retention horizons and back pressure thresholds per layer.

387
00:29:00,040 --> 00:29:05,080
Day 2230, apply purview sensitivity labels at the workspace,

388
00:29:05,080 --> 00:29:09,880
configure one lake item and column level security for sensitive leaves.

389
00:29:09,880 --> 00:29:13,880
Set capacity assignments, dev and prod separate,

390
00:29:13,880 --> 00:29:16,200
and monitor usage and lineage.

391
00:29:16,200 --> 00:29:23,160
Publish the sales model as certified, run demo A and demo B to prove the path and the protection.

392
00:29:23,320 --> 00:29:25,560
Finally, stewardship becomes a living practice.

393
00:29:25,560 --> 00:29:30,200
The sales steward owns schedules, watches lineage, reviews usage weekly,

394
00:29:30,200 --> 00:29:32,520
and prunes, stale branches.

395
00:29:32,520 --> 00:29:36,200
With the fieldpath marked, new makers join without trampling shoots.

396
00:29:36,200 --> 00:29:42,760
The habitat remains calm, the displays stay bright, the migration repeats, without panic.

397
00:29:42,760 --> 00:29:49,560
Quick migration path, from existing data sets to fabric plus direct lake.

398
00:29:49,560 --> 00:29:52,760
Now observe a careful migration, not a stampede.

399
00:29:52,920 --> 00:29:58,120
Many bright feathered flocks already live in older groves, classic data sets,

400
00:29:58,120 --> 00:30:00,440
scheduled imports, brittle gateways.

401
00:30:00,440 --> 00:30:05,080
The path to fabric and direct lake should feel like a calm seasonal move.

402
00:30:05,080 --> 00:30:08,920
Species by species, valley by valley.

403
00:30:08,920 --> 00:30:11,240
First, audit the current clearings.

404
00:30:11,240 --> 00:30:18,040
Catalog every data set, size, refresh cadence, sources, gateway dependencies,

405
00:30:18,040 --> 00:30:21,320
RLS rules, and report attachments.

406
00:30:21,800 --> 00:30:23,480
Lineage becomes your map.

407
00:30:23,480 --> 00:30:29,480
Which reports, drink from which models, which models nibble from which spread sheets or databases.

408
00:30:29,480 --> 00:30:31,480
Mark candidates for direct lake.

409
00:30:31,480 --> 00:30:36,040
Star-like models with fact dimension shape, heavy import refresh pain,

410
00:30:36,040 --> 00:30:39,240
and sources you can land as delta in one lake.

411
00:30:39,240 --> 00:30:41,400
Second, prepare the landing shore.

412
00:30:41,400 --> 00:30:45,960
Create a lake house or warehouse in the correct domain and workspace.

413
00:30:45,960 --> 00:30:49,480
Dev first, capacity assigned, naming aligned.

414
00:30:49,640 --> 00:30:54,360
If you intend direct lake, ensure the tables will arrive in delta format.

415
00:30:54,360 --> 00:31:00,440
Short cuts can bridge endorsed sources, mirroring can bring operational changes downstream.

416
00:31:00,440 --> 00:31:04,200
If M transformations in the old model are intricate or opaque,

417
00:31:04,200 --> 00:31:06,680
lift them into data flows gen 2.

418
00:31:06,680 --> 00:31:08,520
That's where shaping belongs.

419
00:31:08,520 --> 00:31:10,360
The herd will thank you later.

420
00:31:10,360 --> 00:31:13,080
Third, move the water, not the buckets.

421
00:31:13,080 --> 00:31:17,240
Land data into bronze, promote to silver with conformance.

422
00:31:17,560 --> 00:31:21,720
Data types, surrogate keys, business logic defined ones,

423
00:31:21,720 --> 00:31:25,320
validate record counts and joins against the old model.

424
00:31:25,320 --> 00:31:28,520
Use small sample slices to confirm calculations.

425
00:31:28,520 --> 00:31:32,760
When the silver pool looks right, you're ready to teach the species a new song.

426
00:31:32,760 --> 00:31:35,720
Fourth, recreate the semantic language.

427
00:31:35,720 --> 00:31:41,480
Either auto generate a semantic model from the lake house or recreate it by design.

428
00:31:41,480 --> 00:31:45,800
Aligning tables, relationships, and base measures.

429
00:31:46,120 --> 00:31:51,240
Keep names, formats, and folder structures familiar to reduce startle.

430
00:31:51,240 --> 00:31:54,920
Re-implement row level security in the model.

431
00:31:54,920 --> 00:31:56,680
It never migrates by itself.

432
00:31:56,680 --> 00:32:02,840
Test rolls with real identities, a report that looks right to an admin can still frighten a regional bird.

433
00:32:02,840 --> 00:32:05,480
Fifth, reattach the displays.

434
00:32:05,480 --> 00:32:08,920
Point existing reports to the new semantic model.

435
00:32:08,920 --> 00:32:15,080
In many cases, the visuals lift easily if field names and measure signatures match.

436
00:32:15,400 --> 00:32:21,800
Where visuals relied on hidden transformations, replace them with proper measures or silver logic.

437
00:32:21,800 --> 00:32:25,800
This is a gentle pruning, fewer wild vines, more sturdy branches.

438
00:32:25,800 --> 00:32:28,360
Sixth, exercise the new grays.

439
00:32:28,360 --> 00:32:30,920
Validate performance in direct lake.

440
00:32:30,920 --> 00:32:35,000
Filter responsiveness, visual load, bookmark jumps.

441
00:32:35,000 --> 00:32:39,560
If a table is too wide, trim unused columns from facts.

442
00:32:39,560 --> 00:32:43,080
If relationships are confused, simplify cardinalities.

443
00:32:43,080 --> 00:32:48,760
Remember, direct lake sings when delta is present, the model is tidy, and the star is clear.

444
00:32:48,760 --> 00:32:52,280
Seventh, migrate the habitat's roles.

445
00:32:52,280 --> 00:32:55,720
Replicate workspace permissions, certify the model.

446
00:32:55,720 --> 00:33:00,280
Apply purview sensitivity labels at the workspace so tags travel into office.

447
00:33:00,280 --> 00:33:04,040
Configure one lake item and column level controls where needed.

448
00:33:04,040 --> 00:33:09,080
The herd should feel the same boundaries, just stronger, quieter, and native to the habitat.

449
00:33:10,040 --> 00:33:13,400
Eighth, plan the cutover, communicate the window.

450
00:33:13,400 --> 00:33:16,280
Freeze edits on the old model.

451
00:33:16,280 --> 00:33:19,240
Switch report bindings in a controlled sequence.

452
00:33:19,240 --> 00:33:23,880
Keep the old watering hole read only for a short overlap.

453
00:33:23,880 --> 00:33:26,280
A safety pool for comparisons.

454
00:33:26,280 --> 00:33:28,840
Then drain it deliberately.

455
00:33:28,840 --> 00:33:33,240
Deprecate, archive, and remove gateways no longer needed.

456
00:33:33,240 --> 00:33:37,560
The noise of pumps and hoses fades, the river runs on its own.

457
00:33:37,560 --> 00:33:42,040
Ninth, monitor the new trails, watch lineage, usage, and refresh health.

458
00:33:42,040 --> 00:33:48,680
Listen for confused calls, report comments, support channels, and respond quickly.

459
00:33:48,680 --> 00:33:52,840
Small adjustments early prevent flock-wide anxiety.

460
00:33:52,840 --> 00:33:59,320
Celebrate winds, reduced refresh time, faster load, fewer failures, the herd learns the new pattern.

461
00:33:59,320 --> 00:34:01,640
Three quiet warnings from the tree line.

462
00:34:01,640 --> 00:34:03,720
Direct lake requires delta.

463
00:34:03,720 --> 00:34:05,880
Parkay without delta won't do.

464
00:34:06,360 --> 00:34:12,200
Very complex M living inside old data sets belongs in data flows, Gen 2 or pipelines,

465
00:34:12,200 --> 00:34:13,720
not in the birds throat.

466
00:34:13,720 --> 00:34:16,760
An RLS never migrates automatically.

467
00:34:16,760 --> 00:34:19,720
Read defined it in the new model with care.

468
00:34:19,720 --> 00:34:22,440
Strategy governs pace.

469
00:34:22,440 --> 00:34:26,680
Start with low risk species, department sandboxes.

470
00:34:26,680 --> 00:34:30,680
Read only dashboards before touching apex models.

471
00:34:30,680 --> 00:34:34,120
Migrate in bands, not in a single great flight.

472
00:34:34,680 --> 00:34:38,200
Each successful move teaches stewards and makers how to adapt.

473
00:34:38,200 --> 00:34:44,520
That is how an ecosystem evolves, not with panic, but with practiced seasonal grace.

474
00:34:44,520 --> 00:34:51,320
Becoming a steward of the fabric ecosystem, and so the lesson settles softly over the valley.

475
00:34:51,320 --> 00:34:57,560
Power BI thrives when the ecosystem upstream is clear, governed, and alive.

476
00:34:57,560 --> 00:35:03,960
One lake steady beneath, domains tending their territories, and the shared language carrying true.

477
00:35:04,600 --> 00:35:07,240
Continue to observe this ecosystem with care.

478
00:35:07,240 --> 00:35:10,920
Adopt the 30-day baseline, run the field demos,

479
00:35:10,920 --> 00:35:14,520
and begin the sales migration with one calm stream.

480
00:35:14,520 --> 00:35:17,640
For a full system walkthrough and live Q&A,

481
00:35:17,640 --> 00:35:20,600
subscribe and step into the next clearing.

482
00:35:20,600 --> 00:35:24,680
The steward chip never ends and the habitat grows stronger when you return.