This is a set of corrected, extensive notes on a briefing Wikibon received from the head of product marketing for QlikTech, a provider of third-party BI software designed to be used by non-technical business managers directly. The briefing covered the features of QlikView 11, the latest version of the vendor's product, and related issues. Attendees were:
- Jeff Kelly, Wikibon
- David Floyer, CTO, Wikibon
- Bert Latamore, Wikibon
- Jeff Boehm, Vice President of Global Product Marketing, QlikTech
JB: I'm planning to review the key new capabilities in QlikView 11 that we announced today.
QlikView 11 builds on our position in the marketplace in business discovery. We are bringing BI to a broader group of business users, with more self-service, & oriented around business discovery apps that remix & reassemble data from different parts of your organization to answer specific business queries in a more social & collaborative way.
When we thought about Qlikview 11 we went back to our overall mission: Simplifying decisions for everyone, everywhere & making the process of data discovery and decision-making easier. How do people make decisions? For a couple of decades the BI industry has focused on helping people make better decisions based on access to data. Our approach with business discovery simplifies that & makes it available to a broader audience.
But ultimately data alone is not enough. Often when people see data and get ideas from looking at data, they then want to collaborate to get other people's perspectives & opinions on that data. So other people are a second key cog in the decision-making wheel. Today people use stand-alone collaboration technologies, but we believe BI & collaboration need to be tied more closely.
And then the third element is the place where I am. So if I work for a retailer & am on the shop floor, I may see customer behavior around a new display that can influence my decisions on packaging or promotion. That behavior might not be reflected in data because it may not be in the sales log yet. But if I have that insight into behavior on the shop floor, in the hospital, in the factory, wherever I am, that combined with the data & the other people I work with helps me make more effective decisions. We believe those decisions are based on multiple sources of insight, and that is what QlikView 11 is focused on, helping people combine multiple sources of insight.
QlikView 11 is based on five key themes oriented toward those multiple sources if insight:
- The biggest step forward for QlikView 11 is in social business discovery. That is really bringing together the BI capabilities with collaboration platforms. Ultimately we are making the creation, design, development & deployment of analytic apps more collaborative. Also collaborating during the moment of discovery. As you look at information, you want to ask other people questions or share insights with them & explore that data together to gain shared insights and make it easier to bring additional points-of-view on what you are seeing in the application. This process of combining BI & collaboration can be seen in two lights. One is to add social and collaborative capabilities into the QlikView environment. The other is to add Qlikview to a social/collaborative environment such as a Microsoft SharePoint portal or SalesForce Chatter. In QlikView 11 we added new social & collaborative capabilities. These add to capabilities already in earlier versions, such as the ability to share bookmarks of views via IM technologies, and the option to integrate with SharePoint, Chatter, & other collaborative technologies. QlikView 11 adds 2 new capabilities.
- One is the ability to add & share annotations that can become a decision record. I can share a comment or see comments from other people. So when I click on a particular object, I see just the notes made on that object. And I can see what data the commentator was looking at when s/he wrote the comment. Async dialog in which I can then add a comment to that discussion.
- The other new social feature is collaborative sessions, which lets me e-mail a link to what I am seeing to other users, who then can both see and interact with the application in real time, get direct input and real-time dialog about an insight from a particular person. Collaborative sessions allow me to send a link to what I am seeing to other users who can see what I see and interact in real time with me. So several people can share the application, change displays, and discuss what we are looking at in real time. Every person in the meeting can do anything that the originator can do. [call is interrupted, dial back in]
- Comparative analysis:… So in this way as a business user I can see what custom features I want and see how those overlap, and I can perform calculations between them, etc. So these three features – annotations, collaborative sessions, and comparative analysis – are the three biggest new features in this release.
- Mobile business discovery: We have an HTML5 or browser-view approach to extending Qlikview to any mobile device. Initially we focused on iOS and on developing a strong solution for iPad allowing you to have the exact same application running on your desktop & iPad. Over the summer we extended that to support all major tablets & optimize the experience for smart phones. Rather than trying to squeeze an entire dashboard into a 3” smartphone display, it will show one chart or graph at a time and let me interact with that graph or table.
- Rapid Analytic App Capabilities make it easy to deploy QlikView apps out to business users. Unlike traditional BI where you spend a lot of time in initial design & development, our process is typically more agile, an iterative process where you quickly start showing some views of data, get comments & improve it. This latest version improves that process & extends the kinds of applications you can build. For larger customers with large teams developing QlikView apps, we can now integrate with version control systems like Microsoft Team Foundation Server. We've added more meta-elements and new layout & chart controls to provide more flexibility on what you can create & deploy. The other key area is extensibility. We announced integration with Informatica, allowing them to use an optimized load format directly into QlikView. We've expanded on extension objects to add new types of display & data elements such as Google Maps into QlikView. We continue to add new capabilities there to make it easier to develop those objects in Microsoft Visual Studio. We have now added the ability to build extensions that add new calculation capabilities. We have added new types of chart controls such as the ability to see only the top five performers. We added the ability to turn on or off dimensions or expressions. We made it easier to navigate through large applications with tabs on the top.
JK: Can you talk more about how you created the integration with Informatica & others & your architecture under the covers?
JB: Behind the scenes is the QlikView engine, our proprietary in-memory engine that is probably half of the secret sauce behind the product. It's a hi-performance, in-memory engine with a high degree of data compression, unique abilities to create calculations on the fly & cache data based on usage. For simple deployments all you need is that engine, which can run on your desktop. For enterprise deployments you would run that engine on a Windows server either in your enterprise on in the cloud such as on Amazon EC2. You can write a script to have it load data from various sources – flat files, ODBC, SAP, Salesforce. We also have a proprietary high-speed load file format called QVX. Informatica wrote an extension to EPL to write QVX files directly to feed any data it can access into the QlikView engine.
JK: How about handling unstructured data. Is that on your roadmap?
JB: We do have customers who use QlikView with Hadoop & other big data sources. Right now the process is to skim the data you need analyzed and pull that into the QlikView engine. With HIVE you can access data with an SQL-like interface or use a product like Informatica to pull data out of various data sources.
JK: The QlikView engine sounds like a combination of in-memory & disk based storage. What are the storage limitations?
JB: It is an entirely in memory engine, but it can be clustered. We have very large deployments in financial services, public sector, retail, where we handle very large volumes of data in a clustered server environment.
Enterprise Platform: We've made many improvements in how we handle very large clustered deployments & in performance & user experience in that environment. At a high level it is about providing that agility & flexibility while ensuring IT retains governance to maintain control over the data. V 11 has made several improvements. We have the notion of the publisher who creates and publishes a QlikView application to different servers or areas. V 11 makes dramatic improvements to performance and in load-balancing and clustering applications, so I can determine which QlikView apps I want to run on specific servers or how I want to direct traffic based on CPU or size of application and in general give the admin more control over large deployments to ensure consistent performance & user experience. At same time we improved the management console where you administer clusters & manage licenses. We have a visual management console interface & a complete API to let you use third-party control systems or programatically control the environment.
There is the idea of the access point where users get into their QlikView apps, and the end-user interface that is part of the server where I can see all my apps and choose which ones I want to open based on the type of application or individual choices or when they were last updated, etc.
Then we made enhancements around security, adding more document-level auditing to track use of different documents to ensure security compliance and provide more granularity around what features are available on a document or user level. So for instance if I decide that a document should not be shared under any circumstances, I can turn off the sharing features on that application. Or I can allow some users to do collaboration while others cannot on a document.
DF: What is the range of server size or amount of data you can support?
JB: We have deployments with 10s to 100s of Tbytes of data. Part of this is the source-side data vs. the data in QlikView. There is a 90% compression when you bring data into QlikView. Also there is type of data and user volume. So scalability grows in different ways. We have a scalability document that gives you a lot of that information.
JK: What kind of customer environments are you going into? Are you trying to displace BusinessObjects in SAP shops?
JB: It's a very broad range. Certainly we have lots of SAP & Salesforce customers. It is less of a direct replacement play. We find that a lot of people aren't using BI today because the tools are too complex, or require too many levels of approval for licenses, or are too expensive. We are finding that business users are finding and downloading QlikView for free and playing with it & then using it in a department or workgroup, & it expands from there. This happens in a wide range of companies from small retailers to large financial houses and public sector institutions. As QlikView use grows, use of traditional BI tools wanes, but it is a gradual transition. Compared to those traditional BI tools, we're solving a slightly different problem of giving business users the flexibility & access they need.
JK: You are often mentioned with Tableau. Are you coming up against them in bake-offs, & how do you differentiate from them?
JB: Tableau is absolutely in this area also. Gartner calls this part of the BI market “data discovery” and says we own nearly half of that market, but tools like Tableau and SpotFire are there as well. We do see Tableau, although not as much as we see Cognos, Business Objects & MicroStrategies. They offer compelling visualizations & tools for visualization experts. We offer compelling visualizations as well. We don't see them as much where we deploy to large groups of business users who are not technical or not sophisticated in using visualization tools but want that easy interactivity & flexibility we provide. We do see them at the outset when companies are defining their requirements for discovery vs visualization.
DF: Who are your main partners in this area?
JB: Our main partners span 3 categories. We have a very large network of distribution partners, more than 1,200 around the world who sell QlikView both into geographies where we lack direct presence & markets we don't serve in the U.S. and Western Europe. We also have a large group of OEM partners who embed QlikView in their products in specific industries or application areas. We also have large network of technology partnerships with other vendors that provide back-end data warehousing or that provide add-on capabilities, such as Informatica.
We have both a technology development & business partnership with Informatica. We co-developed the integration and have a co-marketing strategy, promoting each others' products in selective opportunities. It is not a resale or OEM opportunity at this time.