High Value Business Intelligence for IBM Platform compute environments
Post on 13-Apr-2017
292 Views
Preview:
Transcript
© 2015 IBM Corporation1 © 2015 IBM Corporation
High Value Business Intelligence for IBM Platform compute environments
© 2015 IBM Corporation2 © 2015 IBM Corporation
IBM Platform Analytics - OverviewArchitectureDemonstration
© 2015 IBM Corporation3
High Value Big Data Business Intelligence for IBM Platform compute environments
Real Time
Historic
What is IBM Platform Analytics?
Provides insight into the data center
Collects a vast array of transactional level data
Processes data at transactional level – retain granularity
Merge business and external data to increase value
Highly extendable, scalable and infinitely flexible
Workload
BusinessUse
Cost
Utilization
© 2015 IBM Corporation4
Why is it important to have insight into these environments?
1. Costs - HPC environments are expensive Purchase Costs of Compute, Storage, Networking, … Data centre space, power and cooling Staff Administration Application Licenses
2. Operational Challenges - Capacity, Configuration, Expenditure, Planning How are resources being utilized? How much resource is required for different projects and applications? Where are the bottlenecks? Where is the spare capacity? What additional resources do we need to meet business requirements? Application Performance?
3. Business Challenges - Productivity, Costs, Time to Market, Competitive Edge How is the HPC environment being used by each part of the business? Are high value projects getting the resources they require? What is the true cost of compute of the business projects? How much does it cost in terms of compute resource and financial costs to design product X? How much cost is required to meet our future business challenges
© 2015 IBM Corporation5
Generic Customers - Insight Requirements
Capacity Planning
Historical workload analysis for better planning
Provide evidence to support future spend
Size license pools effectively
Better negotiate licensing costs
Cluster Configuration
Is workload executing in a timely fashion
Can a change in IBM Platform LSF policies improve my throughput
Is my IBM Platform Symphony resource plan configured correctly
Workload Analysis
How much workload is running through my cluster
Which users and projects use the most resources
Workload profiling
Internal Charge Back
Charge for slot usage
Charge for memory usage
Charge for license usage
Cost of idle capacity
Efficiency
Workload failures
Application efficiency
© 2015 IBM Corporation6 © 2015 IBM Corporation
IBM Platform Analytics - OverviewArchitectureDemonstration
© 2015 IBM Corporation7
Data and Scalability Challenges
BIG DATA
How to collect and store this data?
How to query this data? How to report on this data How to get value from this
data?
Complex workloads
Large user base
Many applications
Multiple geographic locations
Large workload throughput
High cost to run the environmentsBusiness data
Business specific reporting requirementsReal time analysis
Business billing
Thousands of hosts
Multiple clusters
© 2015 IBM Corporation8
How IBM Platform Analytics Works
Raw metrics are loaded from IBM Platform LSF cluster and license servers
Raw data is transformed into a rich report data BI visualizations provide interactive dashboard to
analyze the data Automation controls the processing of the data,
management of historic data and scheduling of static reports
An Extensible solution allows adding of external data to enhance capabilities
Better workload policies Ensure high value workload gets the resources it
requires Optimize resource levels to reduce costs i.e.
Commercial Licenses Identify opportunities to pool and better share
resources Identify workload failure trends Identify efficient and in efficient workload Charge Back modelling
© 2015 IBM Corporation9
Column Based Analytics Database is tuned to support loading and ad hoc query of huge amounts of data:-X86 Based – No proprietary hardware requiredHorizontally scalable – more data, add more nodes
ETL processes data into data rich ready Report Tables ROLAP BI Technology so dashboards interface directly with the data
source (no OLAP cubes) Easy to build and customize reports (quickly) - Every customer has their
own required view of the world Automatically process business level value to reports Framework to collect “whatever” from “wherever”
How IBM Platform Analytics Deals with Volume and Flexibility
DATA VOLUMEFlexibility
© 2015 IBM Corporation10
Cluster 1
Platform Analytics Database
N+1 Technology
Platform Analytics Data Collection Platform Analytics Storage
IBM Platform Analytics Server
IBM Platform Analytics Architecture
© 2015 IBM Corporation11 © 2015 IBM Corporation
IBM Platform Analytics - OverviewArchitectureDemonstration
© 2015 IBM Corporation12
Select Dashboard
Date / Time Filters
Dimension Filters
Interactive worksheets – Click and Filter
Report Parameters
Drill up / down through data
Workbook Overview
© 2015 IBM Corporation13
Report Solutions
Workload accounting
Workload statistics
Cluster capacity
Hardware
Memory requested vs. used
Job statistics
Pending reasons
FlexLM license usage
FlexLM denials
FNM license usage
FNM license denials
FNM workload accounting
Chargeback
Business
© 2015 IBM Corporation14
IBM Platform Analytics Summary
Platform Analytics
• Flexible and scalable high-value Business Intelligence Tool
• Tightly integrated with IBM Platform LSF, IBM Platform Symphony products
• Real time, historic and transactional level data
Why Platform Analytics?
Identify, report and validate resource allocationsQuantify workload and performance
Identify failures and inefficienciesCalculate true cost of compute for applications
Merge with business dataBuild charge back models
Reduce costs Better utilize resourcesImprove efficiency Chargeback
© 2015 IBM Corporation15
Thank You!
Visit ibm.com/platformcomputing
top related