Case Study : MVNO MVNE Data Warehouse Transatel (2008 version)

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    3 Favorites

    Case Study : MVNO MVNE Data Warehouse Transatel (2008 version) - Presentation Transcript

    1. Case study : MVNO / MVNE Data Warehouse Seminar for EFREI engineering school Nov 2008 Mathieu DESPRIEE – BI Manager [email_address] www.transatel.com
    2. Agenda
      • Part 1 – MVNO / MVNE / Transatel
      • Part 2 – Data-warehouse & Business Intelligence :
      • Emergence of the need
      • Part 3 – Business Intelligence :
      • Tools & concepts
      • Part 4 – Project organization
    3. Part 1 MVNO / MVNE / Transatel
    4. Introduction / Definitions
      • MVNO : Mobile Virtual Network Operator
        • An MVNO is a Service Provider offering a mobile service under its own brand without having a mobile radio network
      • GSM operators are referred to as MNO : Mobile Network Operator
        • An MNO has a license on radio frequency ranges (GSM, UMTS…), and all the radio infrastructure
      • MVNE : MVNO-Enabler
        • An MVNE is a service company delivering tools & services to companies wishing to market their services over a mobile network, and thus becoming MVNOs
    5. MVNO and MVNE responsibilities Radio Network Core Network OSS BSS Customer Service Life-cycle Management Marketing Distribution
      • GSM
      • GPRS
      • UMTS…
      • HLR
      • SMS-C
      • MSC
      • GGSN
      • Interco …
      • Service activation
      • CDR collect
      • CRM
      • Billing
      • Processes
      • Operations
      • Call-centre
      • Support…
      • Brand
      • Offer definition
      • Communicat°
      • Marketing mix
      • Sales
      • Distribution network animation
      MVNO responsibility MNO responsibility MVNE capabilities
    6. Transatel Company introduction
    7. Transatel
      • Transatel is an European MVNE (Mobile Virtual Network Enabler)
        • Founded in August 2000
        • Backed with institutional funds such as Natexis and Credit Agricole
        • 80+ employees
        • HQ in France with subsidiaries in UK, Belgium, Netherlands and Luxembourg
      • Serving MVNOs & GSM operators
        • Established brands and businesses: turn key solution to become MVNO
        • GSM operators: infrastructure & services to easily integrate new MVNO and/or build specific co-branded offer
        • A innovative approach to mobile service differentiation and USP
      • Team & Infrastructure
        • Dedicated team with know-how (business, regulatory, technical dev & op.)
        • IT & Telecom platforms
      • Several MVNO launches in several countries
        • Over 200k managed subscribers
    8. Transatel skills & operational capabilities
      • Telecom Business Plan set-up: Consulting
        • Mobile offer definition & positioning, business plan
        • Interco model / cost structure / pricing
        • Regulatory issues
      • Relationship with GSM operators (MNO):
        • Airtime procurement in selected European countries
        • Support in negotiation
      • Mobile Service management: Development and Operations
        • Order entry / Activation / Back office / Customer Management / Call centre
        • CDR mediation / Real-time IN control / Billing (post-paid & prepaid)
        • Fraud detection / Money collection / Bad debt Management
      • Value added services: Development & Operations (incl. « live » prototyping)
        • SIM cards management, personalisation and procurement
        • Voicemail, Unified Messaging, IVR, Follow-me
        • WISP over GPRS
        • Fixed / mobile convergence – Voice / data convergence – GSM / Wifi (VoIP convergence)
        • M2M secured connectivity
      • Pan-European footprint with centralised & mutualised platform
      Turn key solution for MVNOs in Europe from Marketing & Regulatory support to Technical Implementation & Operations
    9. Service portfolio & Operations capabilities Customer Acquisition & Order Entry Provisioning (activations, portability) Billing Customer Care & CRM Value added services SIM cards Order-entry tool web enabled for end-users, distributors Prov. & workflow MNO, service platforms, billing system CRM system web enabled for call-centre & self care Prepaid reload management Prepaid engine IN based Post-perso process for small batches SIM profile definition for large batches SIM mngt & OTA server SIM applets Customisation & branding, call mngt, security Switch & Service node UMS, follow-me, call termination, SMS-GW WISP APN for user authentication & mobile portal mngt + WAP gtway + MMS + remote handset configuration HLR / GMSC (roadmap) Marketing / Business / Regulatory support (offer definition, economic modelling, business processes, regulatory requirements) Logistics (3 rd party) Call centre (3 rd party) Post-paid, invoicing, balance management Payment Collection Revenue Assurance
    10. Transatel interconnection with GSM operators
    11. Transatel MVNE operating models MVNO TRANSATEL MNO End-User MNO End-User MVNO MVNE - ASP E X E M P L E Aggregator TRANSATEL
    12. Vision of a possible efficient market structure Mass market MVNOs Segmented / niche MVNOs Ethnic Sports SME Switchless resellers MVNA ( Aggregator ) M2M Media Distributor Fixed telco MVNE MVNE MVNE MVNO mass market MNO MVNO mass market MVNO niche MVNO niche MVNO niche MVNO niche MVNO niche MVNO niche MVNO niche MVNO niche MVNO niche MVNO niche
    13. EasyBorder : Transatel offer for frequent travellers & cross- border population (FR, BE, NL, LU)
      • Dedicated offer for high roamers
        • 30% to 100% price reduction on roaming
        • Better continuity & dedicated services
      • Commercialised in FR , BE , NL , LUX
      • Niche marketing
        • Direct marketing
        • Flyers & billboards in int’l train stations
        • Advertising in targeted newspapers & websites
        • Active referral program
        • Partnership with frequent traveller programs
      • Sales: telesales & web
      www.transatel.com
    14. Eurokeitai (post-paid) – France www.eurokeitai.com
    15. Transatel international reach Roam Triple Wireless
    16. IT systems to support CRM & billing Billing Customer Management Provisioning CRM (sales & mkt) Order-entry / Customer service Real-time rating IN call-control CDR mediation Prepaid top-up mngmt Post-paid mngmt & invoicing WWW IVR SMS PDF email WWW print-shop Bank interface Reporting & data w/h Customers Distributors MVNO back-office IT interface INAP / CAMEL CDR files from 3 rd parties GSM network Other P/F & systems MNO interface IN P/F Service P/F 3 rd parties, etc. Web Front End (O/E & C/S) API Account Status & Network features Trouble ticketing Subscribers & Catalogue Resources (SIM & MSISDN)  Accounting reports
    17. Transatel Service infrastructure
    18. Part 2 Data-warehouse & Business Intelligence : Emergence of the need
    19. Reporting needs
      • Information is one of the biggest asset of the enterprise. Business users NEED to analyze that data.
      • Finances :
        • Analysis of revenues, costs, profitability, customer/supplier accounting, human resources, financial audits…
      • Revenue assurance
        • Analysis of costs and revenues, detection of revenue leaks
        • Fraud detection
      • Sales & marketing :
        • Sales analysis, orders evolution, customer bills, CRM reports, marketing channels efficiency, offers appropriateness, calls analysis (destination, duration, distribution along time)
      • CRM & Customer support :
        • Statistics on incidents, delay of incident resolution, hotline statistics…
      • Production
        • SLA control. Ex : 99% of billing data processed in 24h
      • CEO, Board
        • Overall performance dashboards, KPI = Key Performance Indicators…
    20. Reporting needs are …
      • Repetitive
        • Finances and marketing will share lots of requirements, but expressed differently
      • Subtly different
        • Finances would like a report starting on Jan, 1 st .
        • Marketing would like the same report, starting on Dec, 10 th (Christmas offer…)
      • Complex
        • Marketing would like to know the revenues generated by an offer (billing data), but with differentiation of the different acquisition channels (CRM data)
      • Expressed by business users
        • Business vocabulary : they won’t use the name of table and columns !
        • They don’t care about the naming conventions in data-models
        • They don’t care about technical constraints
    21. First approach
      • Custom SQL queries
      • Queries against the different databases of the different applications
      • Use Excel sheets (with embedded queries) to federate data, and produce charts
      • Use existing applications to deduce stats (from displayed grids in a search screen for instance…)
    22. Enterprise IT « in real life »
      • Many systems
      • Many data sources
      • Many different data models
      • Different references
      • Duplicates
      •  Problems in data consolidation
    23. Complexity
      • Queries are complex
        • Samples :
          • mvno_activation_monitoring.sql
          • mvno_dump_accounts.sql
          • mvno_financial_report.sql
          • process_monitoring.sql
      • This kind of queries :
        • Use many tables, sometimes from several databases, sometimes from several systems
        • Needs many indexes
        • Produce large result sets
        • Are long to execute
    24. Volumes
      • 1 MVNO = 10,000 to 100,000 subscribers
      • Around 200 CDRs (call data records) per subscriber per month
        • (Up to 800 CDRs per month for some specific offers with a lot of data traffic)
      • 100,000 subs * 200 CDRs * 12 months = 240 M CDRs / year
      • 800 bytes of storage per CDR (including indexes) =~ 190 GB of storage per MVNO per year
      •  doing a SELECT query over the whole data of the year (for instance, sum of all traffic) would require a lot of time of execution, and a lot of resources (disk, CPU) on production server
      • 1 MVNO = 5 tariff-plans = 10,000 rates implemented in billing
      • Call destination analysis tables (numbering plans) = 600,000 destinations in the world
    25. No reactivity to change
      • One report = one query  catastrophic !!
      • Each set of reports evolution needs a project organization :
        • specification by business,
        • technical specification (complexity to interpret business requirements)
        • testing,
        • move to production
      • Even for minor changes !!
    26. OLTP models are not adapted
      • OLTP is the common modeling for business applications
        • transaction-oriented
        • data is highly normalized (cf. 3 normal forms)
      • OLTP models are not designed for large queries
        • Performance impact, database locking
      • OLTP models are not designed to maintain history
        • Often, OLTP systems store a snapshot of reality. For instance, they store the current status of some entity, but not all the history of statuses for that entity and the dates of each change.
        • Keeping all activity for a long time in OLTP systems would have an important impact :
          • Difficult to administrate indexes
          • Performance impact
          • Time to backup & restore database (because systems fail one day or another)
      • OLTP models are not adapted to reporting
        • Data normalization decompose data into many tables and joins, resulting in complex queries
    27. The result :
      • 90% of time spent in developing and producing the reports, 10% spent to analyze them
      • Use of user-interfaces not designed for reporting
      • Too few people know how to produce the key reports
      • Performance impact on production systems
      • Reports are incomplete, no sufficient history
      • “ I have data everywhere, and sometimes I forgot some data-source in my reports”
      • “ I get data from many different systems to fill-in my Excel sheet”
      • “ I’m not totally confident in my reports”
      • “ I make a lot of hypotheses in my simulation, but I cannot verify them”
    28. All these problems = the field of BI
      • Difficulties in data acquisition
        • Need dedicated tools to collect data from different systems, with cleaning and consolidation capabilities
      • Difficulties in data storage with OLTP on production systems
        • Need different data-models to store history
        • Need different databases to store history
        • Need different engine to do the analysis and calculations
      • Difficulties in data restitution with SQL
        • Need tools usable by business users
      • BI = Business Intelligence = Field of technologies that facilitate acquisition, analysis and restitution of data to help in taking decisions
    29. Part 3 Business Intelligence : Tools & concepts
    30. Business Intelligence tools & concepts Production systems Other sources CRM Billing Activation systems Other DBs… Files ETL DataWarehouse OLAP cubes Business Users Querying & Reporting tools Acquisition Storage / Archiving Restitution Datamarts
    31. Business Intelligence tools & concepts (2)
      • ETL
        • Extract : from various production data stores
        • Transform : transform data to match target data-model, merge the different sources
        • Load : in a target database (the data-warehouse)
      • Data Warehouse
        • Repository of the whole enterprise’s historical data
        • Subject oriented : business data
        • Time variant : every change is dated
        • Non-volatile : read-only, never deleted nor changed once committed
        • Integrated : related data from different production systems are gathered
      • Datamart
        • A small repository dedicated to a business subject.
          • Example : one datamart dedicated to cost &revenues analysis, another dedicated to traffic analysis.
        • In the case of Transatel, our data-warehouse is simply constituted of a set of datamarts.
        • OLAP cubes : explained next slide…
      • Reporting tools
        • Querying tools that can be used by business users, non computer specialists (such as excel !)
        • Reports, dashboards
    32. OLAP & Datamarts
    33. Datamart creation : the OLAP approach
      • Reorganize data in an hypercube
        • N-dimensional cubes
      • Facts : the data to be aggregated
        • Data quantifying the business activity (facts are also called “measures” in microsoft tools)
        • Price, revenue, cost, duration of calls, size of data packets, currencies exchange rates, number of visits on web-site, number of incidents…
        • Aggregation operations = sum, average, min, max…
      • Dimensions : the information structuring the facts
        • Time, time, time !
        • Customers, offers, currencies, countries, type of charge, type of call (SMS, voice, data), destinations of calls, marketing channels, …
        • Hierarchies :
          • Time : Year > Quarter > Month > Day
          • Geography : Zone > Country > Region > City
          • Organization : Business Unit > Line of business > Market
      Time Geography Offers Cells contains facts data
    34. OLAP : On-Line Analytical Processing
      • Cubes are the way OLAP technology organizes summary data into multidimensional structures.
      • Dimensions and their hierarchical levels reflect the queries that can be asked to the cube.
      • Aggregations are precalculated and stored in the multidimensional structure in cells at coordinates specified by the dimensions.
      • Sophisticated algorithms in modern OLAP engines minimize storage
      • Calculating all possible aggregations would require substantial processing time and storage, which is unnecessary. As a result, there is a tradeoff between storage requirements and the percentage of possible aggregations that are precalculated.
      • Different OLAP storage modes for different needs :
        • MOLAP : multidimensional OLAP : fastest calculation, larger storage, not updated real-time
        • ROLAP : relational OLAP : smaller storage overhead, slowest calculation, real-time
        • HOLAP : hybrid mode
      • OLAP engines use specific querying language (MDX)
        • OLAP client tools generate MDX queries
        • MDX can also be written by developers for advanced querying
    35. OK, I need a cube, where do I start ?
      • The star data-model (or snowflake)
      Facts Dimension Dimension Dimension Dimension
    36. Dimensions hierarchies
      • Example of hierarchy for a time dimension
      • The cube is configured so that the OLAP engine knows the hierarchy structure, and can precalculate aggregates at different dimension levels
    37. Real world example 1
      • Cube dedicated to revenue analysis from post-paid billing
        • Advanced point :
        • Transatel billing produces bills for customers in several countries, and in several currencies.
        • Currency is a special point of interest, because we can’t sum prices in different currencies.
        • Currency conversion rate variations along time are stored in a dedicated fact table
        • A “reporting currency” dimension has been created
        • The business user chooses in which currency he wants the aggregates
        • Cube has been configured with advanced calculation rules to help in correct processing
    38. Real world example 2 : cube for billed charges
      • The datamodel (snowflake)
    39. The cube structure
    40. The cube structure Main Facts (charges) Measures : charge, sum minutes, sum SMS, sum MBytes …
    41. The cube structure Measure : Subscriber count
    42. The cube structure Measure : {subscriber x month} counting an artificial measure to ease some computations conceptually it’s a crossjoin, in this case, it’s better to create it from the datamodel, with a dedicated SQL query
    43. The cube structure Measure : the currency exchange rate .. to display the data in €, even if the bill is in another currency (conversion is done by an advanced MDX script)
    44. The cube structure Dimensions the Date dimension is used several times (role-playing dimension)
    45. The cube structure Relationships a « regular » relationship ie : a BilledCharge has a DateKey for the BillingDate dimension
    46. The cube structure Relationships a « fact » relationship ie : the same table (subscriber) is used as a dimension (organizing attributes) and as a fact (for counting)
    47. The cube structure Relationships a « referenced » relationship ie : the Country dimension is not directly related to facts. We have to go through another dimension (customer) to get it
    48. The cube structure Relationships a « many-to-many » relationship ie : subscriber measures (count) are not related to products. We have to go through the billed charges to know that. This is a many-to-many join : Dim Product  Fact Charges Fact Charges  Dim Subscriber Dim Subscriber  Fact Subscriber
    49. ETL
    50. ETL
      • Extract
        • Multi-source
        • Batch oriented : usually ETL batches are ran when load is low on production systems (by night)
        • Connectivity to different technologies, to various DBRMS implementations
      • Transform
        • Manipulate complex data
        • Merge from different sources
        • Implement business rules.
        • Cleanup, sorting, deduplication…
        • Optimized for performance and large volumes
      • Load
        • Load data in fact tables and in dimension tables
        • Produce logs to files or database for traceability
      • Others
        • Controls execution flows : manage dependencies between tasks, control execution, manage errors, logs
    51. Examples of transformation available in SSIS
      • Fuzzy Lookup Transformation
        • Look up values in a reference table using a fuzzy match
        • Example : the field “how did you heard about Transatel” is a free-text field on the website. The data contains a lot of variants for same words (“friend”, “a friend”, “freind”, etc). The fuzzy logic can help in merging all the variants
      • Slowly Changing Dimension Transformation
        • Help in managing a the update of a slowly changing dimension
        • Example : Information about the customers change continuously, but we want to keep some history.
          • A change in customer address  we don’t keep the old value (override)
          • A change in company VAT Number  we ‘duplicate’ the customer entry, to keep old financial data aligned with the old VAT Number (for annual reports)
      This is called a surrogate key
    52. ETL : control the execution !
      • Data is very sensitive, quality of reports rely on the quality of data
      • Loading a datawarehouse can imply very complicated rules on extraction, transfer and transformations
      • Business users won’t trust a datawarehouse, if they’re not confident on how it has been loaded.
      • So, one of the key aspects in ETL process implementation is the control of how runs the data transfer and transformation
      • ETL solutions come with a lot of tools to control process execution, and with alerting systems (eg: mail sent to the DBA) to raise the error.
      • Microsoft SSIS organize ETL packages in control-flow and data-flows.
        • Cf next slide
    53. ETL package example in SSIS
      • Control flow and data flow
    54. Restitution & Reporting
    55. Reports topology Dashboards Adhoc reports Mass reporting executives operational teams
      • KPIs, trends
      • 5% of users
      • data and trends analysis
      • need of interactivity
      • need of drill-down, drill-through
      • 20% of users
      • static reports designed for specific purposes
      • 75% of users
    56. Reporting : OLAP cube browsing
      • Users can browse data through pivot charts
        • Colums and rows are taken amongst available dimensions
        • Content of pivot-chart is measures (attributes of facts)
      • Excel is a good low-end tool for cube browsing
      • Demo
    57. Reporting : Control access to data
      • Security issues :
        • Security of access and segregation of data is important : anybody may not access to any data.
        • Especially in the case where reports are produced for external partners / customers : they can connect to a secured web-site and download reports they’re interested in.
        • Some business data may also be confidential, and destined to top-executives only.
        • Reporting tools and OLAP engines provide the ability to define roles and permission on data, at every level (dimension, attribute of dimension, cell, etc.)
      • Simplification for business users :
        • Cubes may contains a lot of dimensions, dimension attributes, and facts. In some tools, it’s possible to define a subset of a cube, and restrain some users to access only at this part of data (“perspectives” in microsoft tools)
    58. Part 4 Project organization
    59. Collect the business needs
      • The best way is to start from existing reports, and existing excel sheets frequently used by managers & executives
      • Organize reporting needs by functional domain, and if possible by data-source
      • Define at a macro level the model you need
        • Determine dimensions
        • Determine facts
      • Look at data access permissions
    60. What users want
    61. What they specify
      • “ Put all information you have, we will select things afterwards”
    62. What you’ll manage to do
    63. What they really need ( thanks to Yves Cointrelle for the joke idea)
    64. Select requirements and organize phases
      • Talk with business users, again, again and again !
      • Organize requirements in a matrix of choice :
      • Eliminate complex and non-urgent needs, and organize project phases
      complexity business priority P1 P2 P3
    65. Define granularity of data
      • A radar diagram can help you to :
      • Identify the needed dimensions and their hierarchies
      • Decide the level of granularity you need to keep in your datamarts
    66. Learnings from experience…
      • Dimension design : beware of cardinality !
      • Example : Time dimension
      • First design : one table for all granularities, from year to second
      • 365*24*3600 = up to 31’536’000 rows by year !!!
      • This design was catastrophic on performance of everything : loading, pre-processing, querying, custom calculations (advanced MDX)
      • In fact, bad analysis of business need
        • Day granularity was enough for most of time dimension usage (date of bill-cycle, date of call, date of subscriber creation, etc.) = 365 rows per year
        • Second granularity was useful in a unique case : time of call placement, only 86400 rows and rarely used.
      • 2 different time dimensions : a “Date” dimension (used a lot), and a “Time” dimension (used once in the model)
      • Result : Pre-processing time improved by a factor of 20…
    67. Learnings from experience…
      • ETL flows design
      • 6 first months of loading of the datawarehouse after each bill-cycle : the ETL flows NEVER finished successfully !
      • There are always little problems or unexpected inconsistencies in data
      • Restarting a whole ETL process is long and costly for hardware resources
      • Don’t hesitate to separate the 3 parts of E-T-L
      • Design your flows so you can restart a phase without restarting everything, using intermediary storage if necessary
    68. Datamart development methodology
      • Be AGILE !
      • Estimate the quality of data
      • Estimate constraints on data (ETL rules)
      • Design core part of the data model
      • Show to the users what is possible
      • Help users to define what they need
      • Do a prototype if necessary (with manually loaded data)
      • Implement ETL flows, rules, and data clean-up
      • Implement the data model and the cubes
      • Produce the first reports
      • Control the results with pre-existing reports
      • Discuss the first results and problems with users
      • Adapt and improve ETL rules and flows
      • Improve cube model to adapt to users comments
      • Add the advanced computed indicators (MDX scripts)
      • Refine the aggregation designs to optimize cube performance from real usage statistics
    69. Project methodology
      • Start small !!
      • Build progressively, use iterative approach with frequent delivery to production
        • Benefits : give a quick feedback to business users
        • Validate the technical architecture
      • Organize a weekly project committee
      • Involve final users
      • Get trained
        • Developers should be trained specifically on the chosen BI tools
        • Business users have to be trained to use the new tools (and avoid a reject)
      • Buy consulting from specialists of necessary
    70. Issues not to under-estimate
      • It can take some time to stabilize the flows and the rules in the ETL
        • With some applications, data is never totally clean.
        • You may need additional rules or even some manual processing from time to time
      • You may spend some time in operations with the first versions of your cubes
        • To adapt the aggregation configuration from the real usage statistics
        • To adapt the indicators you compute to what the users want
      • You need to spend time in the modeling phase
        • With a bad model, cube may be technically impossible to optimize. You can even crash you server with a bad query
        • With a bad model, you can obtain false calculations
      • MDX is a very difficult language
      • Business users will become greedy of your reports !
        • One day behind schedule to produce a sensitive report, and all the management will be on your back !!
        • One false statistic about the business, and everybody will hate you … Do a lot of testing to stay confident !
    71. Some vendors of BI software…
      • ETL
        • Informatica
        • IBM
        • SAS
        • Cognos
        • DataMirror
        • Microsoft SSIS
        • Sunopsis
        • Business Objects
        • Oracle
        • Open Source : Talend, Enhydra
      • Storage
        • Generalists : Oracle, Microsoft SQL Server
        • Specialized :
          • Red brick
          • Teradata
          • Sybase IQ
          • SAS
          • Netazza
          • Datallegro
        • Open Source : MySQL
      • OLAP
        • Microsoft Analysis Services
        • Hyperion
        • Oracle
        • SAP BW (dedicated to SAP solutions)
        • SAS
      • Reporting
        • Business Object / Crystal Report
        • Cognos
        • Hummingbird
        • Hyperion Solutions
        • Microsoft reporting services
        • Information Builders
        • Oracle
        • SAS
        • Informatica
        • MicroStrategy
        • Actuate
        • Open Source : JasperReports, Pentaho
    72. Bibliography
      • Yves Cointrelle
        • BI training session – Homsys Group
      • Miscrosoft MSDN for Analysis Services
        • http://msdn2.microsoft.com/en-us/sql/Aa336310.aspx
      • Wikipedia
    73. Thank you ! electronic version www.slideshare.net/mathieu.despriee feel free to contact me for any question [email_address]
    74. Jobs and Internships at Transatel

    + mathieu.desprieemathieu.despriee, 2 years ago

    custom

    3550 views, 3 favs, 0 embeds more stats

    Case study : MVNO / MVNE Data Warehouse Seminar for more

    More info about this document

    CC Attribution-NonCommercial-NoDerivs LicenseCC Attribution-NonCommercial-NoDerivs LicenseCC Attribution-NonCommercial-NoDerivs License

    Go to text version

    • Total Views 3550
      • 3550 on SlideShare
      • 0 from embeds
    • Comments 0
    • Favorites 3
    • Downloads 511
    Most viewed embeds

    more

    All embeds

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories