Enkitec-Exadata_OpenHouse2
10
Enkitec Exadata Open House
Transcript of Enkitec-Exadata_OpenHouse2
Enkitec Exadata Open House
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 2
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 3
Why Exadata?
• Reduced Implementa/on Time • Balanced, Scalable Configura/on
Be;er
• Performance, Performance, Performance
Faster
• Consolida/on and Simplifica/on • Lower TCO Across-‐the-‐Board
Lower Cost
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 4
Fast Deployment Example: Direct Marketing Company
Compression Ra/os Compression Type GB Compression
Ratio
None 33.49 0% Advanced Compression 10.51 219% HCC For Query Low 5.05 563% HCC For Query High 3.04 1002% HCC For Archive Low 2.67 1153% HCC For Archive High 2.29 1360%
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
select * from table;
No Compression
OLTP
HCC Qry Low
HCC Qry High
HCC Arc Low
HCC Arc High
• First ¼ Rack Exadata Rack System in Production in about 2 weeks
• Staff DBA created initial database using DBCA
• 3 Production Databases running in 2 months
• Second Exadata ¼ Rack used as Dataguard Physical Standby
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 5
Performance Example: Mid Size Retailer
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 6
Performance Example: Mid Size Retailer
Current Stats: • Currently runs on M5000 with 32 cores
• Test 1: Nightly forecast runs multithreaded 4 stores at a time, and typically takes 2.75 hours to complete
• Test 2: PO Build Plan process which runs approx 90-120 seconds currently
• Test 3: Three ad hoc queries running 56 minutes, 27 minutes, and 4 minutes respectively
Exadata Results: • Tested on ¼ Rack Exadata with 16 cores
• Test 1: Parallel factor of 18 stores at a time, Best forecast run approx 53 minutes for all stores
• Test 2: PO Build Plans completed in approx .2 seconds
• Test 3: Queries completed in 4.5 minutes, 8 minutes, and 5 seconds respectively
• Non-required Indices removed which reduced space by 25% and reduced runtime by approx 30%
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 7
Consolidation Example: Large Direct Marketing Firm
Current Stats: Non-scalable business model: New customers required dedicated hw/sw platforms, slow onboarding, complex, expensive management. Nine current customers required:
• 29 different DB instances • 21 different servers • 3 different OS versions • 7 different versions of Oracle DB • 6 different BI/Analytics apps • 3 different ETL tools • and a partridge in a pear tree
Exadata Results: Consolidation into (2) Full Exadata RACs
• Carved out 1/4 Rack for Dev/Test/UAT • Prod DB instances reduced to 9 • Five “cluster groups” allow resource sharing with segmentation
• Created DB and IORM Resource Management policies to ensure consistent response times
• One version of Oracle DB/OS/HW • One vendor for Support • Centralized management via OEM • Reducing and standardizing on fewer 3rd party apps/tools
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 8
Exadata 5 Year Total Cost of Ownership Comparison
5 Year Comparison (Hardware, Software, Support, Labor, Datacenter)
Traditional Platform $6.3M 2 Exadata 1/4 Racks $3.7M
Savings $2.6M (5yr NPV)
• 41% Savings over Traditional Platform
• Not inclusive of Exadata business operational benefits
Five Year Cost Comparison
$0$200,000$400,000$600,000$800,000
$1,000,000$1,200,000$1,400,000$1,600,000$1,800,000$2,000,000
Year 1 Year 2 Year 3 Year 4 Year 5
An
nu
al N
et E
xpen
se
Traditional Platform Exadata
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 9
Demo Time
12/23/12 © Copyright 2009 Enkitec, Inc. Proprietary and Confidential. 10
Questions / Contact Information
Questions?
Contact Information : Kerry Osborne
Email: [email protected] Blog: kerryosborne.oracle-guy.com
www.enkitec.com
