Inside Oracle Exalytics and Oracle TimesTen for Exalytics - Hotsos 2012

Inside Exalytics and Oracle TimesTen for Exalytics
Mark Rittman, Technical Director, Rittman Mead
Hotsos Symposium 2012, Dallas

T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com

Mark Rittman

• Mark Rittman, Co-Founder of Rittman Mead
• Oracle ACE Director, specialising in Oracle BI&DW
• 14 Years Experience with Oracle Technology
• Regular columnist for Oracle Magazine
• Author of forthcoming Oracle Press book on OBIEE 11g
• Writer for Rittman Mead Blog :
http://www.rittmanmead.com/blog
• Email : mark.rittman@rittmanmead.com
• Twitter : @markrittman

• First time at the Hotsos Symposium ;-)


About Rittman Mead

• Oracle BI and DW platinum partner
• World leading specialist partner for technical excellence, solutions delivery and innovation in
Oracle BI
• Approximately 50 consultants worldwide
• All expert in Oracle BI and DW
• Offices in US (Atlanta), Europe, Australia and India
• Skills in broad range of supporting Oracle tools:
‣ OBIEE
‣ OBIA
‣ ODIEE
‣ Essbase, Oracle OLAP
‣ GoldenGate
‣ Exadata
‣ Endeca


Agenda

• Introduction, and the context for Exalytics
• Oracle Exalytics BI Machine
• Exalytics Under the Covers
• TimesTen for Exalytics
• Summary


What is Oracle Business Intelligence?

• Oracle BI provides a set of web-based tools for reporting and analyzing data
• Based around dashboards, built around graphs, tables, pivots etc
• Multiple server architecture that splits data access / presentation elements
• Based on Siebel Analytics technology
• Connects to multiple data sources
‣ Oracle DB & Exadata
‣ SQL Server, DB/2, Teradata
‣ Essbase, Oracle OLAP, MSAS
‣ ADF View Objects etc
• Now re-engineered as part of FMW11g


Oracle BI within the Oracle BI/DW Technology Stack

• BI provides the query and presentation element for BI/DW
• Catalog of reports, analyses, alerts, KPIs
• Additional calculation and
data federation capabilities Web Server

• Aggregate navigation
• Security, scoping of data
Oracle BI

ERP/Apps DW Essbase

SAN / NAS


How Oracle Business Intelligence 11g Works

• Data sources are mapped into an internal
dimensional model called the Oracle BI Repository
• Users request data through analyses, which
translate into logical SQL requests
• The Oracle BI Server converts these logical
SQL requests into one or more physical
SQL, MDX or other queries
• These are then processed by the underlying
data sources, using their optimizations
• Data is then combined and returned
to users in the form of dashboards


In a BI/DW System, Where Does the Time and Heavy Lifting Go?

• Ideal is to return data to users with consistent sub-second response times
• But large data volumes and unpredictable queries can make this tough
• When analyzing query response times, most of the times goes in
‣ The database returning the query results
‣ The storage tier returning data to the database
‣ The network connecting storage and the database
• Tuning, efficient queries and Exadata can help with this Oracle BI

‣ But queries can still sometimes take >2 mins or so
‣ And query times can be erratic
- Some are instant ERP/Apps DW

- Some take 2-3 minutes
- Depends on load on the database
SAN / NAS


BI System Performance Objectives

• Every query returns data immediately
• No performance penalty for drilling-into and investigating the data
• Response time is consistent
• Response-time stays constant even when you scale up to 100’s of users


Demonstration
OBIEE Dashboard without Exalytics Optimizations


Potential Solutions to Query Performance Issues

• Buy an Exadata box (or several Exadata boxes...)
• Add materialized views, more indexes, in-database OLAP to the DW
• Move data into an OLAP server such as Essbase
• But...
‣ Can be expensive
‣ Adds to DBA complexity
‣ Requires new skills (Essbase etc)
‣ Scaling issues for OLAP servers


Another Solution - In-Memory Databases

• Move some, or all, of the data being queried into RAM
• Sub-second response times
• Consistent response times
• A solution being applied elsewhere for similar problems
Oracle BI

In-Memory
DB/Cache

ERP/Apps DW

SAN / NAS


Oracle Exalytics In-Memory Machine X2-4

• BI Appliance that combines hardware + BI software
• Sun hardware, Four Intel Xeon E7-4800 series processors
• 1TB of RAM, made up of 64 16GB DDR3 ECC registered DIMMs
• QDR (40Gb/second) InfiniBand connectivity to Exadata & Exalogic
• 10Gb and 1Gb Ethernet interfaces
• Optimized version of Oracle BI EE
• In-memory version of Oracle Essbase
• 3.6TB of disk, for internal use
• Integrated Lights-Out Mgmt (ILOM)
• Sold as an appliance/engineered system
‣ Software only available in
combination with hardware


Exalytics as the Exa-Machine for BI

• Runs the BI layer on a high-performance, multi-core, 1TB server
• In-memory cache used to accelerate the BI part of the stack
• If Exadata addresses 80% of the query performance,
Exalytics addresses the remaining 20% Oracle BI

‣ Consistent response times for queries
‣ In-memory caching of aggregates
‣ 40 cores for high concurrency In-Memory DB/Cache
‣ Re-engineered BI and OLAP software
that assumes 40 cores and 1TB RAM

ERP/Apps DW


Enables High-Density Analysis of Big Data

• BI tier is rarely the bottleneck, but it can be if very dense visualizations are used
‣ Sparklines, grid of charts etc
• Exalytics’ 40 cores and 1TB RAM make higher density presentation viable
‣ Single query sent to the database
‣ Exalytics breaks data up to create microcharts
• Also helps support high numbers of concurrent users (100+)


What is Exalytics?

• Hardware Element
‣ Sun Fire X4470 M2 server
‣ 1TB RAM, 40 Cores, 3.6TB HDD
• Software Element
‣ OBIEE 11.1.1.6 with Exalytics Enhancements
‣ Oracle Essbase 11.1.2 with Exalytics Enhancements
‣ Oracle TimesTen 11.2.2.2 for Exalytics
‣ Runs on 64-bit Oracle Linux
(Exalogic distribution)
• OBIEE and Essbase are licensed as
Oracle BI Foundation
• Exalytics features can only be used in
conjunction with Exalytics hardware


Exalytics Architecture

• Developer tools used for defining Developer Workstation

and maintaining aggregate definitions Oracle BI Administration Tool

• Client tools for OBIEE repository,
Summary Advisor Wizard

nqcmd
TimesTen and Oracle Database
TimesTen Client

SQL Developer

Oracle Database 11g Client

Oracle Business Intelligence

• Exalytics layer contains OBIEE,
Oracle TimesTen for Exalytics
WLS Admin
Oracle BI
Server

TimesTen and Essbase
System Comp.
Aggregate
WLS Managed BI Server Schema
(not pictured) Server

• Database tier contains FMW
Database for RCU Schemas
database repository, including Enhanced Usage Tracking
Summary Statistics
usage tracking and summary statistics

Exalytics Under the Covers - How Does it Work?

• Exalytics brings together different technologies, which are still standalone products in
their own right
• To harmonise and optimise their use within Exalytics, it utilises the following techniques:
‣ In-Memory Adaptive Data Mart
‣ In-Memory Intelligent Result Cache
‣ In-Memory Cubes
• Some of these are genuine "secret sauce"
‣ New functionality and algorithms
‣ You can only get them through licensing Exalytics
• Others are descriptions of DW/BI strategies, or existing product functionality, extended
to take advantage of the capacity for processing in memory that Exalytics has


In-Memory Adaptive Data Mart

• Commonly-used aggregates are copied into Oracle TimesTen for Exalytics
• Past query patterns are analyzed and suitable aggregates recommended
• Oracle BI Server then uses these aggregates to make queries run faster
• Aggregates change over time in
response to changes in query patterns TimesTen BI Server
• Tools are provided for managing

Exalytics
and populating these aggregates Aggregates

Data Warehouse
Detail-level
Data


In-Memory Intelligent Results Cache

• OBIEE has an advanced result cache, which can support:
‣ Queries matching previous ones
‣ Queries matching a subset of the results of a cached query resultset
‣ Queries aggregating the results of a cached query resultset
• Stored in cache files on the OBIEE server filesystem
• With Exalytics, the 1TB of RAM typically means
that these files are held in memory
• RAMdisk can also be created for
BI Server & Presentation Server TimesTen BI Server
working area files

Exalytics
In-Memory
Result
Cache


In-Memory Essbase Cubes

• Essbase is being enhanced to provide better in-memory capabilities
• Some features will be common to all Essbase releases, some will require Exalytics
‣ Smart Storage Manager stores frequently access blocks in-memory
‣ In-Memory Cache optimizes block access
‣ Lock Manager improves parallelism


So What is Oracle TimesTen (for Exalytics?)

• TimesTen dates back to 1998, acquired by Oracle in 2005
‣ It is an In-Memory Database
‣ The whole of the database is loaded into memory
‣ It is persisted to disk, for full recovery capabilities
• Originally designed for exceptionally fast and highly concurrent read/write of single rows
of data (airline booking, trading desks, etc)
• Can be deployed in two ways:
‣ Standalone database
‣ In conjunction with Oracle, as part of In Memory Database Cache
• Previous versions of TimesTen were usable with OBIEE, but...
‣ Because of its focus on fast OLTP, was slow for reporting
‣ Limited analytical capabilities


Oracle TimesTen for Exalytics

• New version of TimesTen specifically for Exalytics (and only available with Exalytics)
• Support for analytic functions
‣ Perform all the processing at source
‣ Combine with being in-memory = should be very fast
• Column compression
‣ Whitepaper cites 5x
‣ Given the hardware capacity, we could seriously contemplate loading the whole Data
Warehouse into memory
‣ Opens up lots of interesting design potential
• We can load aggregates into TimesTen, leave base data at source, and use OBIEE’s
Vertical Federation capability to seamlessly report across both
‣ All hidden from the end-user, all they will know is that their reports run fast!


TimesTen and OBIEE Architecture

• Oracle BI Server communicates with TimesTen through TimesTen Client
• Summary Advisor, and nqcmd use Oracle BI Server to access TimesTen
• Typical single TimesTen database per Exalyics machine
‣ Max TimesTen database size around 300MB
- Due to need to set aside equal Summary Oracle BI nqcmd
Advisor Server
Temp size for the Perm size selected
• Clustered Exalytics boxes can be daisy-chained ODBC
together using InfinBand connections Checkpoint
Files
‣ For HA scenarios, does not increase
available RAM TimesTen
Memory-Resident
‣ Summary advisor scripts write to both TimesTen Database

databases, replicating aggregates
‣ TimesTen databases can be “wired together”
for failover/HA purposes
Log
Files


TimesTen Extensions for Exalytics

• Notable new functionality in this version includes:
‣ Columnar Compression
‣ Analytic functions: AVG, SUM, COUNT, MAX, MIN, DENSE_RANK, RANK, ROW_NUMBER,
FIRST_VALUE and LAST_VALUE
‣ Analytic clauses: OVER PARTITION BY and OVER ORDER BY
‣ Grouping clauses: GROUP BY CUBE, GROUP BY ROLLUP, GROUP BY GROUPING
SETS
‣ Grouping functions: GROUP, GROUPING_ID, GROUP_ID
‣ WITH clause
‣ Aggregate expressions over DISTINCT expressions


TimesTen for Exalytics Columnar Compression

• The version of TimesTen for Exalytics features “In-memory columnar compression”
• Compression is specified against individual columns, or groups of columns
• Some will compress better than others
‣ A compression group is where the same compression dictionary is applied to values
in the columns within it
‣ Examples of use would be surname & maiden name, home_city & city_of_birth
• Compressed columns take slightly longer to load, but should:
‣ be faster to query than uncompressed columns
‣ take up less disk space
‣ Compression typically reduces the 3x overhead of usual TT storage (compared to
source data in a file) back to the original dataset size


Enabling Columnar Compression in the Administration Tool

• When initially created, the TimesTen database used by Exalytics is set up without
columnar compression enabled
• To enable, check the Compress Columns checkbox in the Database Features dialog
‣ Also, disable WITH_CLAUSE_SUPPORTED
‣ TimesTen can use them in SELECTS, but not
during data load (disable during loads only?)


Summary Advisor within Oracle BI Administration Tool

• Utility within Oracle BI Administrator tool that recommends aggregates
• Bases recommendations on usage tracking and summary statistics data
• Captured based on past activity
• Runs an iterative algorithm that searches,
each iteration, for the best aggregate

Proposed
Aggregate
Fact Table
20 rows
Row Count

X =
Predicted
Previous
Query Time
30 secs New Query 0.06 secs
Time
Exisitng
Base Fact
Table
100000 rows
Row Count


Underlying Enhanced Usage Tracking Data

• Historically, usage tracking has been tracked using S_NQ_ACCT
‣ Holds basic usage tracking statistics + logical SQL query
• Now supplemented by S_NQ_DB_ACCT
‣ Extra usage tracking information, includes physical SQL
• Exalytics Summary Advisor uses S_NQ_SUMMARY_ADVISOR
‣ Contains summary statistics, execution time etc
‣ Gathered at same time as usage tracking when
Exalytics is enabed
‣ Contents can be derived from usage tracking if needed


Required Configuration Steps for Exalytics/Summary Advisor

• Configure TimesTen DSN for in-memory aggregates
• Configure opmn.xml in middleware home for TIMESTEN_DLL
• Create TimesTen database, user account and permissions
4. Create sample table in TT database using SQL*Developer
• Import or create the TT table in RPD
• Add usage tracking and summary statistics tables to RPD
• Enable usage tracking via EM/MBean
• Restart system

Assumes system has been enabled
for Exalytics by Oracle ACS or Partner


Demonstration
The Oracle Exalytics Summary Advisor


Creating In-Memory Aggs Step 1 : Select Facts and Time Scope

• Select Tools > Utilities > Summary Advisor
• Select facts from the BMM layer to consider for the exercise
‣ Can also generate stats from usage tracking data if none exists (upgrade etc)
‣ And can load parameters from a file
• Select start and end date for wizard to consider (optional)


Creating In-Memory Aggregates Step 2 : Set Min Query Time, Target

• Select minimum elapsed time for queries, to consider for exercise
• Select physical database details for TT database


Creating In-Memory Aggregates Step 3 : Set Script Target Location

• Summary Advisor will generate a script that has to be run through nqcmd.exe
‣ Same type of script as generated by Aggregate Persistence Wizard
• Script can be re-run to refresh aggregates periodically
‣ Drops and recreates aggregates, no incremental refresh


Creating In-Memory Aggregates Step 4 : Define Stopping Criteria

• Select maximum run-time for summary advisor (0=unlimited time)
‣ Useful when lots of summary statistics data to process
• Set limit for how little each additional aggregate improves performance
‣ Stops lots of little aggregates that barely improve performance being recommended
‣ Ideally, set to around 10% for good balance
between good recommendations and
good dashboard performance


Creating In-Memory Aggregates Step 5 : Run + Filter Recommends

• Aggregate recommendations can then be generated
• Filter page lets you select which ones to implement
‣ Suggested approach is to add one aggregate at a time, test and implement
‣ For smaller, more tested RPDs you can select all if required


Creating In-Memory Aggregates Step 6 : Generate Aggregates

• Wizard then creates the script, which can be run through the nqcmd utility
‣ Command-line access to the BI Server
• Run command either at the developer workstation, or on the Exalytics server
• Re-run at any time to refresh aggregate tables with new data

cd [obiee_client_home]oraclebiorahome
bifoundationserverbinnqcmd.exe
-d exalytics -u biadmin -p welcome1
-s c:summary_advisor_scriptsagg_wiz.sql


What Does the Summary Advisor Create?

• Aggregate tables in the TimesTen database
‣ One per aggregate fact table
‣ One per shared dimension
• Physical tables in the Oracle BI Repository physical layer
• Logical Table Sources in the BMM layer
‣ All mapped in and active


Oracle BI Server then Redirects Queries to Aggregates

• When users then query the dashboard, queries will get redirected to the
TimesTen tables
• Aggregate navigation is additive, so single aggregates should satisfy other queries

TimesTen BI Server

Exalytics
Sending query to database named
TT_AGGR_STORE, Aggregates
connection pool named TT_AGGR_STORE,
logical request hash 9b1a12f5, physical
request hash 2f1461f5: [[select
sum(T44087.Billed_Qua0000A5C2) as c1,
sum(T44087.Revenue0000A5BC) as c2from
ag_2123023723 T44087
Data Warehouse
Detail-level
Data


Summary Advisor Usage Recommendations & Limitations

• Start by generating and testing single aggregates each time
‣ If one aggregate creation fails, whole process fails; tricky to “un-pick” process
‣ Focus on a few facts, recover any errors, then move to next facts
• Constraint time window to a representative range
• Be aware of how best to use the repository and source data
‣ Run “estimate levels” prior to Summary Advisor, helps to ensure unique level keys
‣ Do not run against value-based hierarchies, or skip-level/ragged (not supported)
‣ Cleanse data before running Summary Advisor - ensure unique level keys, no
NULLs in dimension key references
‣ Implement indices on the source systems


Alternative Use Case : Manually Load “Hot” Data into TimesTen

• Summary Advisor only loads recommended aggregates into TimesTen
‣ Usually a good approach, as based on historic usage and expected gain
• It is possible to manually load source data into TimesTen
‣ Aggregates using Aggregate Persistence Wizard
‣ ETL using ODI
‣ Replication using Golden Gate
• Aggregates would get automatically mapped in RPD
• ETL’d data would need manual mapping
‣ Replacement LTSs
‣ Fragmentation
‣ Additional LTSs
• Remember TT RAM restriction:
maximum of around 300GB


Troubleshooting and Diagnostics

• Enable detailed logging for summary advisor scripts
‣ Session variable in RPD: LOGLEVEL >=4
‣ Prefix aggregate statements
- set variable LOGLEVEL=4 : delete aggregates
- set variable LOGLEVEL=4 : create aggregates
• Logs in nqquery.log, Errors in nqserver.log
• Enable detailed logging for the Admin tool
‣ [instances]instance1configOracleBIServerComponentcoreapplication_obis1l
ogconfig.xml
‣ <Level>
<IncidentError>1</IncidentError>
<Error>1</Error>
<Warning>1</Warning>
<Trace>32</Trace>
</Level>


How Does it Perform?

• Have performed some software-only testing, but Exalytics box not yet available
• Performed some standalone testing of TimesTen, and also compared TimesTen
queries to standard queries
• Difficult to judge without
‣ Running on an Exalytics box - BI Server has optimizations for this hardware
‣ Representative data volumes
• Keep an eye on our blog for findings once the box is delivered
‣ http://www.rittmanmead.com/blog


Demonstration
OBIEE Dashboard following Exalytics Optimizations


In-Memory Aggregate Caching Benefits & Limitations

• For individual dashboards, analyses etc, caching works well
• For general aggregate re-usability, depends on rest of model, query types etc
‣ Use in conjunction with results caching; similar benefits and limitations
• Caching (of aggregates) makes TimesTen not suitable for real-time (OLTP) sources
• Needs to be used in conjunction with database and storage tuning
‣ Ideal link is with Exadata (uses InfinBand)


Summary

• Well it certainly looks interesting...!
• Exalytics’ 40 core server with 1TB of RAM provides a very fast hardware platform
• Addresses the “mid-tier” performance issue for BI systems
• Infiniband connnection makes integration with Exadata particularly effectivee
• TimesTen for Exalytics provides the in-memory database cache
• Tools provided as part of Exalytics make it possible to intelligently populate the cache
• Benefits for Essbase users in-terms of specific optimizations for Essbase
• Looking forward now to getting our hands on the hardware
‣ Contact us for a Proof of Concept exercise, using our “Exa-lab” in Brighton, UK


More Information

• Thank you for attending this presentation
• More information can be found at http://www.rittmanmead.com
• Contact us at info@rittmanmead.com or mark.rittman@rittmanmead.com
• Look out for our book, “Oracle Business Intelligence Developers Guide” due Q3 2012
• Follow-us on Twitter (@rittmanmead) or Facebook (facebook.com/rittmanmead)


Inside Oracle Exalytics and Oracle TimesTen for Exalytics - Hotsos 2012

Recommended

More Related Content

What's hot (20)

Similar to Inside Oracle Exalytics and Oracle TimesTen for Exalytics - Hotsos 2012 (20)

More from Mark Rittman (20)

Recently uploaded (20)

Inside Oracle Exalytics and Oracle TimesTen for Exalytics - Hotsos 2012