2 BSO 3 ASO 4 Hybrid 5 Performance Demo 6 QampA 7 About Brian Marshall VP of Delivery 20 years IT and EPMBI Experience 15 years with USAnalytics 150 projects with USAnalytics Presented at ID: 934287
Download Presentation The PPT/PDF document "AGENDA Introductions 1 Essbase Storage T..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Slide2AGENDA
Introductions
1
Essbase Storage Types
2
BSO
3
ASO
4
Hybrid
5
Performance Demo
6
Q&A
7
Slide3About Brian Marshall
VP of Delivery
20+ years IT and EPM/BI Experience15+ years with US-Analytics150+ projects with US-AnalyticsPresented at Kscope every year since 2010
Frequent blogger at EPMMarshall.com
Slide4About US-Analytics
Dallas-based, Hyperion-focused for over 18 years with continuous business growth
We are nimble and respond quickly to customers’ needs
Over
500
clients and over 1,000 successful Hyperion engagements
Seasoned business and technical acumen with EPM and BI initiatives
Over 65 professionals with 12+ years each of Hyperion experience and certifications
Active leaders in the Oracle community
Founder of Hyperion Professional Women's Forum, advisory board leadership, conference presentations, webinars, EPM Speaker of the Year at Kaleidoscope in 2015 and 2014
Corporate culture of integrity with 100% customer commitment
Managed servicesManaged services team is Dallas based, each with 10+ years of experience
Proven processes for all aspects of managed services
Slide5ABOUT US-ANALYTICS
Managed Services
Upgrades & Migrations
Implementations
Infrastructure
Process & Advisory
Services
Big Data
Data Governance
Business Intelligence
Financial Close & Consolidation
Planning & Forecasting
Solutions
Data Integration
Training
Accolades
Original Oracle Hyperion and Pillar Partner
Oracle Hyperion Financial Management 11
Oracle Hyperion Planning 11
Oracle Essbase 11
Oracle Data Relationship Management 11
2013, 2014, 2015
2015 Oracle TOLA
EPM Partner of the Year
Slide6Essbase Storage Types
BSO
The original Essbase technology
Great for procedural calculationsEveryone knows and loves BSOASOHas been around for ages, but still far less adoptedCan do some procedural calculation, but struggles with upper level detailFantastic for large data setsHybridNew in 11.1.3.500…but more of a beta
Much better in 11.1.2.4 and OACGreat for Planning, but not supported
Slide7Overview
BSO
Data stored in blocks and indexed
Block size matters, although not as much as it used toHas gotten better and better for performance in many ways, but lagged behind in other waysMakes great use of memory and CPU powerStruggles to make use of even lightning fast storage solutions
Far and away the most adopted Essbase data storage typeStruggles with end-user performance with attribute dimensionsGets far worse with bigger sparse dimensionsGets even worse than that when users stack attributes
Slide8Use Cases
BSO
Still the easiest storage method to support
No shortage of information and support on the internet and in the communityOracle support at least seems to know something about BSOTraditional Planning models
Basic procedural calculationsLimited need for attribute dimensionsAverage-sized dimensionsExisting BSO applications that don’t have issuesThere are a ton of these…if its not broken, don’t go looking for a reason to fix it
Slide9Overview
ASO
Data stored in bitmaps
Block size doesn’t existNo more dense and sparseInstead, think stored and dynamicCommon Misconceptions
Shared members requires a dynamic hierarchyFormulas require that an entire dimension be dynamicRealityShared members are allowed in a stored hierarchyMultiple Hierarchies enabled allows us to have our cake…and eat it tooGreat for very large data sets (more on this later)
Attribute dimension performance is fantastic, even with stacking
Slide10Use Cases
ASO
Not as easy to support as BSO, but still far easier than Hybrid
Still no shortage of information and support on the internet and in the communityOracle support seems to know far less about ASO and escalation doesn’t seem to help that much
Traditional Rack-n-Stack modelsNo need for procedural calculationsJust aggs and basic dynamic calculationsModels with very large dimensionsModels with extremely large data setsModels that make use of many attribute dimensions
Existing BSO applications that do have issuesTiming of data refreshesAttribute performance
Slide11Overview
Hybrid
In many ways, the holy grail
A BSO cube that I don’t have to aggStores data in blocks and dynamically creates bitmapsBut…there’s no such thing as a free lunch
Much of what we love about BSO procedural calculations sufferPartially hybrid cubes help in some cases, but then we are back to needing aggsAttribute dimension performance is just as terrible as BSOWorks great in Hyperion Planning, but is not technically supportedDoesn’t exist in PBCS, yet, but is on the roadmap
Doesn’t performance tune at all like BSO…or ASO
Slide12Introduction
Performance
2 Applications
EssBench (ASO, BSO, and Hybrid)Account (1025)Entity (8767)Product (8639)
Real-World Hybrid + ASO Reporting CubeAccount (273)Entity (115)Cost Center (3599)Employee (13205)
Slide13Data Loads
Performance
Slide14Aggregations
Performance
Slide15Restructures
Performance
Slide16Block Size in Hybrid
Take everything you know about block size in BSO and throw it out the window
Make the block much smaller than a traditional block
Make the account dimension sparse for exampleHave only one or two dense dimensions that are relatively smallWhy?End-user performance suffers massively with larger block sizeThe entire block is pulled in as it dynamically builds the upper level intersections
Slide17Block Size in Hybrid
But wait…block size matters
So what’s the downside?
Upper-level calculations will be slower as Essbase everything will be calculated as dynamically sparseGenerally, block can have an impact on this depending on your calculationsDense dynamic is considerably faster than sparse dynamic
Slide18Standard End-User Performance
Performance Demo
Slide19Attribute End-User Performance
Performance Demo
Slide20When DOES Hybrid Work?
When you fewer, smaller sparse dimensions
When you don’t need attributes
Great for an existing application that you want to get better end-user performance out of without needing to aggregate all the time
Slide21When DOESN’T Hybrid Work?
More than 2 large sparse dimensions (5000+ members)
Need to use attribute dimensions, especially more than one
If you want support from Oracle if you are using Planning
Slide22So Why Not Always ASO?
ASO seems like its better, faster, and generally more awesome right…not exactly
Procedural calculations
Can be just as fast or faster than BSOFar more limited than BSO around how you handle upper-level membersInfinitely harder to code and maintainExisting application conversion would be expensiveEasier to just add a reporting cube rather than converting an existing cube
Slide23What Should I Use?
ASO, BSO, or Hybrid!
In many cases, Hybrid will give you a great boost in end-user performance by limiting the need for aggregations
But, Hybrid still has issues with what it supports and performance on large sparse dimensionsHow about the best of both worlds?BSO or Hybrid for your main Planning applicationASO reporting cube that takes a live feed for dataSmart PushCustom On-Prem Smart Push (until 11.2 comes out…we hope)
Slide24Q&A
?
Slide25Shameless Plug
Visit my blog:
EPMMarshall.comFormerly HyperionEPM.comVisit my benchmark:EssBench.comConnect to #orclepm
on twitter:
Slide26