Enterprises are rapidly adopting stream computing backbones, in-memory data stores, change data capture, and other low-latency approaches for end-to-end applications. As businesses modernize their data architectures over the next several years, they will begin to evolve toward all-streaming architectures. In this webcast, Wikibon, Attunity, and MemSQL will discuss how enterprise data professionals should migrate their legacy architectures in this direction. They will provide guidance for migrating data lakes, data warehouses, data governance, and transactional databases to support all-streaming architectures for complex cloud and edge applications. They will discuss how this new architecture will drive enterprise strategies for operationalizing artificial intelligence, mobile computing, the Internet of Things, and cloud-native microservices.
Link to the Wikibon report - wikibon.com/wikibons-2018-big-data-analytics-trends-forecast
Link to Attunity Streaming CDC Book Download - http://www.bit.ly/cdcbook
Link to MemSQL's Free Data Pipeline Book - http://go.memsql.com/oreilly-data-pipelines
29. 31
The Digital Enterprise Requires Performance
Fast Queries
Scalable SQL
Real-time dashboards
1 Trillion row scan a second
Scalable User Access
Multi-threaded processing
Converged transactions and analytics
Scale-out for performance
Live Loading
Stream data
ACID transactions
Multiple sources
30. 32
MemSQL Architecture
Operational Database and Data Warehouse in One
Historical Data
Disk-optimized tables
with compression for
fast analytic queries
Live Data
Memory optimized tables
for analyzing real-time
events
Streaming Ingest
Real-time data pipelines
with exactly-once
semantics
32. OLTP Sources DashboardsData Integration Data Mart
SAP on Oracle
SQL Server
SAP Data Services MySQL
Tableau
Case Study: Slow Performing Analytics with Batch ETL
33. OLTP Sources DashboardsData Integration Data Mart
SAP on Oracle
SQL Server
SAP Data Services MySQL
Challenges:
• 24 hour ETL process
• Minimize operational
database impact
Tableau
Case Study: Slow Performing Analytics with Batch ETL
34. Case Study: Slow Performing Analytics with Batch ETL
OLTP Sources DashboardsData Integration Data Mart
SAP on Oracle
SQL Server
SAP Data Services MySQL
Challenges:
• 24 hour ETL process
• Single threaded
operation
Tableau
Challenge:
Slow running queries
35. Case Study: Faster Analytics with Live Loading
OLTP Sources Dashboards
SAP on Oracle
SQL Server
Tableau
Why MemSQL?
- Fast distributed data ingestion
- 22 hrs ingestion to 43 minutes
SAP Data Services
36. Case Study: Faster Analytics with Live Loading
OLTP Sources Dashboards
SAP on Oracle
SQL Server
Tableau
SAP Data Services
Why MemSQL?
- Fast queries with no pre-
aggregates
- 20x dashboard acceleration
37. +
RESULTS
39
30x faster
Data latency
dropped from
22 hours
43 minutes
80x query response
improvement over
mySQL
Profitability analysis
of customer
logistics was
transformed from
weekly to daily
38. 40
Accelerating the frequency and accuracy of fraud detection on live
operational systems with streaming architectures
+
US Oil and Gas
Exploration Company
39. 41
Case Study: Invoice Analysis Performance on Oracle
Oracle
Step 1
Vendor Invoice
committed to
Oracle
Financials
Step 2
Analyst checks
for duplication
and invoice
accuracy
Step 3
Batch rule job
runs at night
comparing all
invoices for
duplication
Step 4
Invoice cleared for
payment. The compare
job fails often enough
and invoice has to be
paid out. For Duplicates,
clawback needs to
happen.
ERP Source Invoice Audit Reports
40. 42
Case Study: Invoice Analysis Performance on Oracle
Oracle
Step 1
Vendor Invoice
committed to
Oracle
Financials
Step 2
Analyst checks
for duplication
and invoice
accuracy
Step 3
Batch rule job
runs at night
comparing all
invoices for
duplication
Step 4
Invoice cleared for
payment. The compare
job fails often enough
and invoice has to be
paid out. For Duplicates,
clawback needs to
happen.
Challenges:
- Batch job took 12 hrs to run on
Oracle
- Oracle tuning unable to perform
- Unable to migrate from Oracle
due to core operational system
ERP Source Invoice Audit Reports
41. 43
Case Study: Invoice Analysis Performance on Oracle
ERP Source
Oracle
Step 1
Vendor Invoice
committed to
Oracle
Financials
Step 2
Analyst checks
for duplication
and invoice
accuracy
Step 3
Batch rule job
runs at night
comparing all
invoices for
duplication
Step 4
Invoice cleared for
payment. The compare
job fails often enough
and invoice has to be
paid out. For Duplicates,
clawback needs to
happen.
Challenges:
- Batch job took 12 hrs to run on
Oracle
- Oracle tuning unable to perform
- Unable to migrate from Oracle
due to core operational system
Invoice Audit Reports
Challenge:
- Incorrect invoice payments cost
millions per year
42. 44
Case Study: Continuous Invoice Analysis for Savings
ERP Source
Oracle
Step 3
Analyst checks MemSQL for
duplicate invoices in real-time
Step 2
CDC job replicates
Oracle to MemSQL
Step 4
Faster invoice
clearance with
reduced clawbacks
and duplications
Fraud Reports
Step 1
Vendor Invoice
created in the
system.
Why MemSQL?
• Reduced 12 hr batch process to 22 seconds
• Invoice validated at time of submission
43. 45
Case Study: Continuous Invoice Analysis for Savings
ERP Source
Oracle
Step 3
Analyst checks MemSQL for
duplicate invoices in real-time
Step 2
CDC job replicates
Oracle to MemSQL
Step 4
Faster invoice
clearance with
reduced clawbacks
and duplications
Fraud Reports
Step 1
Vendor Invoice
created in the
system.
Why MemSQL?
• Reduced 12 hr batch process to 22 seconds
• Invoice validated at time of submission
Why MemSQL?
• Dramatic reduction in
clawbacks
• ROI realized in months
44. MemSQL and Attunity – Enabling Digital Business
Operational
Sources
Real-Time
Analytics
Oracle, SQL Server,
SAP, Hana, DB2
Dashboards, Real-Time
Applications
Pipelines Live Data Historical Data
Why MemSQL and Attunity?
• Live synchronization without impacting application performance
• Save time and effort reporting and analyzing directly on transactional data
• Cost effective with enterprise performance and operational capabilities
• Run anywhere spanning on-premises to cloud
• Augment Streaming ingest data with Enterprise ODS/Transactional context
Change Data Capture