From what I’ve read and understood, Project Crescent is best applied to ad-hoc, undirected analysis work rather than directed reports, displaying current operational data activity. This blog aims to understand more about Project Crescent, and when it is best applied to business questions.
From the front-end reporting perspective, Project Crescent is a major new Business Intelligence addition to Microsoft SQL Server Denali. It is primarily designed to assist business users, across an organisation, to ‘storyboard’ their data by facilitating unstructured analyses of the data, and to share it in the Microsoft Office applications that we are familiar with already. Project Crescent has its genesis in the SSRS team, and is focused on allowing business users to analyse, mine and manipulate their data ‘on the fly’. The solution will be browser-based, using Vertipaq and Silverlight to product rapid BI results by allowing users to interact with their data. Further, this analysis will be powered by the Business Intelligence Semantic Model (BISM), and will only be available in Sharepoint mode. This criteria would need to be met in order to use Project Crescent, so this may not work in every environment.
How does this perspective differ from reporting? Well, standard reporting takes business users down a particular structured route; for example, the parameters are pre-prepared in the report in drop-down format, and the dataset is already ‘set’ about the data that can be shown. This offers limited options for business users to interact with the data. On the other hand, Project Crescent will allow unstructured analysis of the data, which does not lead business users down a particular route. Rather than produce reports, the unstructured analysis does not offer a real separation between ‘design’ and ‘preview’. Instead, the users ‘play’ with their data as they go along, ‘thinking’ as fast as they ‘see’.
‘Unstructured’ type of data analysis is great with most data sources. For straightforward operational needs, the structured reporting facillities will answer the immediate, operational business question. This is particularly the case when considering the display of data from an Operational Data Store (ODS). What does the ODS give you? The ODS basically helps support people who conduct business at the operational level, who need the information to react quickly. Essentially, it sits between the source systems and the data warehouse downstream, and the data usually has a short shelf-life. It is an optional aspect of the business intelligence architecture, and worth considering since many organisations have at least one ODS. There are many different definitions for the ODS, but here are some common chacteristics:
– it intakes data from disparate sources, cleanses and integrates them
– it is designed to relieve pressure from source systems
– stores data that isn’t yet loaded into the data warehouse, but is still required now
One example of this ODS deployment is in contact centre management. For example, the contact centre manager will need to know if there is a high abandoned call rate at the contact centre. As a result of this, they may need to temporarily direct calls to another team in order to meet the unusually high demand. To do this, however, they need to be aware that the abandoned call rate has reached a particular threshold. This reporting requirement is operational rather than strategic; a strategic decision may involve using a number of metrics to identify whether some longer-term re-organisation is required, but the ODS helps the business to run optimally here and now.
This leads us to another question; how can we best display the data in the ODS, and display it? Data visualisation of ODS information has very specific requirements since it needs to show, at a glance, how the business is operating. There are a number of different data visualisations which can assist in showing ‘at-a-glance’ information very effectively. There are plenty examples of data visualisations that can assist in the ‘here and now’ business question facilitation. For example, for information on bullet charts to display target information, please see an earlier blog. It is also straightforward to produce KPIs in Reporting Services, and if you’re interested, please see this post by Jessica Moss. and so on. In SQL Server Denali, these visualisations can be produced using SSRS, Sharepoint, Excel, and PowerPivot, not to mention a whole range of other third-party applications such as Tableau. There are more details about the Microsoft-specific technologies in the SQL Server Denali roadmap, which can be found here on TK Anand’s Blog.
To summarise, from what I’ve read so far, Project Crescent is aimed more at ad-hoc analysis work rather than displaying current operational data activity. From the roadmap, I gather that Microsoft SQL Server Denali will have something to please the business users who require ‘unstructured’ data analyses, in addition to the high standard of functionality already available for those who require structured data analyses in response to a changing business environment. I look forward to its release!
HI Jen
Your Crescent posts are really useful.
One thing I can't determine from these posts or the video is whether you can connect to local data. If the app is all browser based, what if I need to do some ad-hoc analysis on a CSV on my hard-drive? Or even a very large local database? I would assume that data would need to be uploaded somewhere.
If done properly, Crescent could be a threat to other rapid-fire tools (Tableau, Spotfire, etc). However, if data needs to be uploaded somewhere, it's one step that would slow down the user, and weaken the benefits.