Jumpstart: MongoDB BI Connector & Tableau

Editor's Notes

• #3: 1:00
• #4: Mission is to help people see and understand data…Big and Small…..to people of all skillsets..regardless of whether they are a data scientist, doctor, teacher...We believe everyone can benefit from harnessing the power of data... And today I will to provide an overview of our Big Data vision and how we execute that vision, especially as it pertains to MongoDB and data of the noSQL... And also impart some awareness of some best practices and techniques customers might want to employas they use Tableau on top of MongoDB for performant and efficient results..
• #5: People who know the data should ask the questions. It seems so simple … BI infrastructure and processes are designed to prevent this.
• #6: Software should be designed for deeper thinking. In fact - Out of way – user flow – prompted, wizards stay focused on tasks -- software should do the heavy lifting for you, not ask you even more questions. Need to Build software - smart decisions and guides you to questions you never thought to ask
• #7: Analytics at scale can drive change. Data is one of the greatest assets. don’t know – advantage Your customer’s organizations need to change to harness the potential of their data. Not one dashboard, everybode - changing processes , relationships , yes, even the power structure Analytics, and the quest for facts can rapidly drive that change.
• #8: Start with sharing how we are thinking/approaching the Big Data Space. In addition to everything in Tableau, focus on cloud, analytics, mobile, we have a this perspective on Big Data. Access to all Data Data of any size, stored in any database or file, or in any format Structured or unstructured, messy or perfectly organized We also believe there is and will always be many sources of data that one person wishes to use Data at Scale Being able to have a fast experience with my data. How do I connect to Billions of records and get to the data point fast. We’re not big advocates of sampling data or using Machine Learning tools. We want users to have access to all their data, at scale, to integrate with other data, and find insights fast. No compromises. Discovery Today, the data needs to be created in Hcatalog for a user to connect and analyze. But what if I want to explore my data and build a schema, or build a table in hcatalog? We’re building the tools to make discovery on unstructured data an easy step for the end-user.
• #9: How do we achieve Analytics for all data… -------------------------------------------------------------------------- Now how does Tableau do this all to make Fast, Easy & Beautiful possible for Big Data? It boils down to 6 main pillars First, we provide broad access to Big Data platforms – We have a number of direct connectors in our product including: Hadoop: Cloudera Impala & Hive, Hortonworks Hive, MapR Hive, Amazon EMR with Impala & Hive, Pivotal HAWQ, IBM BigInsights NoSQL: MarkLogic, Datastax Spark: Apache Spark SQL Cloud: Amazon Redshift, Google BigQuery Operational Data: Splunk We enable visual analytics without coding – Business users can visualize their data using drag-and-drop operations without writing complex SQL, Java code or MapReduce jobs. Tableau simplifies the task of analyzing data - users can discover visual insights about their data faster than they ever could before. We have a hybrid data architecture - Tableau can connect live to data sources or bring it in-memory. Live connectivity works great when connecting to fast interactive query engines such as Impala & Spark and large datasets. However, we can also augment and accelerate slower data sources by creating an extract of the data and bringing it into our in-memory Data Engine. We enable data blending - Distributed Data is often times an even bigger challenge than Big Data. It is rare an analyst’s data is nicely packaged in a single place - instead, data is all over the place residing in disparate technologies and platforms. Tableau enables users to traverse across data sources by blending Big Data with other data sources (e.g. Salesforce, MySQL, Excel files), allowing organizations to keep their data assets where they reside. We invest in our platform query performance - As data volumes grow, Tableau continues to invest in core query performance improvements that help facilitate real-time conversations with data. Features in 9.0 such as parallel queries are especially relevant on distributed scale-out architectures such as Hadoop Speed to insight….is achieved both by speeding access...as well making the whole process of analytics more agile...business users can ask and answer their own questions..without relying on specialists... MongoDB helps provide our customers the same benefit…application layer/business availability layer……Don’t have to ETL or reformat the data…without having to move data…significant cost and complexity…can be eliminated We provide a consistent interface to visualizing data – The way that you use Tableau is nearly identical across all data sources. If you or your other data analysts have used Tableau with other data sources, they’ll be familiar with our visual interface to data which is the same across big and small data. Across the spectrum of “big data” vs. “small data” we can see that the tools we use at each end of the spectrum are different. It’s not viable to use complex, high-end tools for simple problems over a few thousand records; nor is it viable to solve complex queries over billions of records with Excel. However, with Tableau you have a single tool that lets you span the range. Partnering with best of breed.. Analytics for all of your data: Tableau empowers people throughout the organization to answer questions of their data, large or small, in real-time. The more questions they ask, the more value they extract from the data, leading to smarter business decision every day. Tableau works with best of breed technologies and works seamlessly with Big Data databases in addition to more traditional databases, you can have one interface into all of your data. This makes Tableau itself the best “Tableau for Big Data” tool.
• #11: 5:30
• #16: demo1.mov
• #18: MongoSQLD = mediator process = involved in translation to/from SQL Change from BI Connector 1.x (which was a PostgreSQL foreign data wrapper based solution)
• #34: demo2.mov
• #35: Compare to BIC 1.x which used foreign data wrapper & didn’t support the same level of pushdown
• #36: 16:00 Note all Agg Framework tips & tricks apply
• #37: demo2.mov
• #38: 21:00
• #43: demo3.mov
• #44: 25:00
• #46: We’ll talk about a Cold, Warm, Hot strategy for data which is something that has come out of our customers who are using Tableau with petabytes of data.
• #47: and a new way of thinking about how you store, process and use data...Gone are the days, where you used one platform for everything...Paradigm of cold warm and hot data... Not everything needs to be sub 2 second response time...sometimes queries that run for a while are ok...cold use cases...ETL..algorithmic analysis The idea of this is you have a large cold data store of raw or unprepared data, this could be structured or unstructured that is used for data exploration use and query response may be slow. Next is warm data which is data that has been prepared or ETL’d and is typically in a relational datawarehouse or fast analytical database. Finally we have the hot, which will typically be smaller in memory datasets used on the highest value vizzes where performance is critical. In the NoSQL world – where you store your data, how you access your data is a function of what you want to do with it...You can optimize that at many levels... Cold, Warm, Hot represents three tiers of data Each Tier varies by: 1. The amount of data and level of aggregation 2. The speed of access required 3. How frequently you need to access the data
• #48: Just a double click into our Cold Warm Hot Strategy…If you dont like that nomenclature you could also call it...Raw, Entity and Analytical structures... We can use guided analysis methods in Tableau to build out interactive visualisations that leverage the hot tier for the highest value elements while still providing a path to drill down into the granular data when needed. This chart is just a generalization to help understand the different tiers of data and data access The larger the data volumes are the more likely you are to have them stored in the unstructured and semi structured forms....Fly all over US with a TB of storage..infathomable amount of storage not possible a few years ago... MongoDB before the launch of this optimized connector was traditionally “Cold”. You could either be exploring the raw data or working off of a prepared view…. Relational databases can be seen as your “Warm”. This is the data we’re mostly used to. These are the Kimball stars, the cross tabular reports we cut our teeth to…Typically, you are working from prepared views of the data and you need fairly fast access…Semi analytical structures supporting production reporting.. Lastly, you have in-memory Tableau Data Extracts as well as other fast analytical databases (like Vertica/Exasol) that took burden off traditional data warehouses to speed things up for read-style queries…can we have MPP style execution to make our DW and speed to business intelligence faster..., extracts(Columnar) and fast analytical databases are solely used to back your Tableau dashboards…We also have the loads of tools that have cropped up generating loads of SQL queries We’re also seeing Tableau making an investment with our Hyper engine...We want to put more things into Tableau and federate things....bring more data into Tableau and access it faster...
• #49: MongoDB before the launch of this optimized connector was traditionally “Cold”. You could either be exploring the raw data or working off of a prepared view….
• #50: MongoDB is a great place to capture and store all of your information. There are a couple main use cases that I hear about from customers: 1. In the Data Exploration/Mining use case, customers will simply use Tableau as a way to understand general trends and outliers of the data. These are your core data scientists, who'll use Python, Scala, Java to process data and use Tableau to give you quick insights into your data….. In the 2nd, They’ll do Ad-Hoc Report Conceptual Modeling where they will ask….hey is this idea even worth exploring....theyll use Tableau to figure out what parts of the data they actually want to use in preparing a more finalized view of the data…so theyll explore concepts and migrated them into more of an aggregated or analytically optimized structure or up the stack into an Impala to expose it to business users.. Technologies for Hive on MapReduce include Cloudera, Hortonworks, MapR, Amazon EMR to do these types of use cases..
• #51: Here are some use cases I’ve seen out in the field speaking with customers.. Fixed width financial transactions, storing raw text, Airline booking records, green screen style printouts Ecommerce...Dumb amounts of data, in massively multiplayer online games... E Commerce...All browsing activity, click activity, people like to do AB testing, campaign optimization to deliver target ads Telco...So many different devices that have to be in place in the infrastructure to move communications forward...Every vendor of even hardware makes a big data play...then you have the service providers...the ATT and the Verizons of the world..
• #52: 30:00
• #55: Mention: Wisdom’s talk Pre-empt first question about other BI Tools / ODBC / MySQL clients