Business dashboards are big, and my aim is to create a
stunning one that combines my twin passions for HTML5 and Windows Azure. In
this series I’ll be sharing my progress on iterative development of an HTML5
dashboard hosted in Microsoft’s Windows Azure
cloud that works on computers, tablets, and phones. Here in Part 1 I’ll share
initial thoughts on design and a glimpse of our first prototype,
themed for a fictional company named Fabrikam Imports. In subsequent posts I’ll
have additional companies set up with a variety of themes and content.
Goals
Here are my goals for the dashboard, which will be realized
in steps:
Compelling. I want
to provide a compelling user experience with animation and transitions. It will
take some refinement to find the balance between an interesting level of
movement and being overly flashy or distracting.
Branded. Since the dashboard will most
often be used by businesses, it will need to carry their branding. This can be
done overtly with a corporate logo and reinforced with a background , such
as the bamboo image we’re using for Fabrikam Imports.
Broad Reach. The dashboard
needs to be accessible across computers, tablets, and phones and work well
whether accessed by mouse or touch. This means fluid layout that accommodates
different form factors and orientations, and interactive elements large enough
to support touch.
Useful. The
dashboard is more than eye candy: it also needs to be a valuable source of
aggregated business information. Which means it had better be able to integrate
to many kinds of data sources.
Content Versatility.
Charts aren’t the only content to be shown. I’d ultimately like to also support
tables, maps, media, and other content. In some cases we’ll provide more than
one view of the same content.
Data Source Versatility.
Supporting the most popular formats and access methods should make it easy
to integrate dashboards to many kinds of data sources without a lot of rocket
science. I initially plan to support data in XML and JSON format data that can
be retrieved from cloud blob storage, web services, or feeds.
Explorable. While
I’m not trying to compete with BI offerings, I would like to have
more than one dimension to the data. When viewing a chart of data, you should
be able to select a value of interest and drill down a level if the data allows
that.
Lightweight and Deluxe
Editions. For my dashboard, I’m envisioning a lightweight edition and a deluxe
edition. I’ll explain the differences later in this post.
Fast Setup. The
lightweight version should allow you to set up a decent dashboard in just
an hour or two if you have your data available in XML or JSON format. Setup
time for the deluxe implementation will depend on how ambitious you are and will typically involve some custom integration work.
The Front
End
If you’ve been following this blog, you’ve noticed that many
of my recent experiments in HTML5 have been around controls such as a counter,
bar
chart, and ticker.
That’s all been driving toward creation of a dashboard. But creating all the
controls I need on my own (and taking them to commercial grade) would take a
long time, so the best approach will be a flexible dashboard that can adapt to
include controls from a variety of sources. With this approach, I can combine
best of breed controls with my own. For charting, I’m currently trying out JSCharts. So far I’ve found JSCharts to be
easy to use and it displays very efficiently. I am encountering some display
issues however after resizing so the jury is still out on whether I’ll be using
it in the long term. Either way, I eventually want to support multiple chart packages.
The current layout has a corporate logo and a ticker at top (which
can be paused/resumed with a click or touch). Below this I have three views in
mind: list, tiles, and zoom.
List, Tiles, and Zoom Views
List View. In list
view, you’ll see a list of content that’s available. You can check an item to
control whether it is shown or hidden in other views, and you can select an
item to go directly to it in zoom view.
Tiles View. In
tiles view, you’ll see all of the content (except items you’re chosen to hide)
shown in fluid layout to fit your device. Content buttons allow you to take
actions on the data, such as zooming in for a detail view.
Zoom View. In zoom
view you’re focused on one item of content and it fills the full window. Here
you might be able to see the content in more than one way, such as a table and
chart side-by-side. It’s also from here you may be able to explore some data
sets by drilling down.
Implementation technologies for the front end are HTML5,
CSS3, JavaScript, and JQuery. Content controls will be a mix of my HTML5 controls
and public/commercial controls.
The Back
End
The back end runs in the Windows Azure cloud. I mentioned
earlier I have two editions in mind, a lightweight edition and a deluxe
edition.
Lightweight Edition
The lightweight edition is meant to be fast and simple to
set up and incredibly cheap to operate. It runs completely out of Windows Azure
blob storage and that’s also where the data is kept. This will generally cost
just a few bucks a month to operate. Accessing the dashboard URL in a browser
transfers the front end code down to run in the web client. Ajax calls issued
by the web client pick up data files from blob storage (which are in XML or
JSON format) and the dashboard content is populated. To update the data in the
dashboard, you push out new data in the form of XML or JSON files to blob
storage. This can be done with small ‘update agent’ programs that integrate to
your internal systems or manually. This model does have some limitations, namely
authentication. The best you can do to secure this is to generate shared access
signatures (secretive URLs) for the blob container. On the plus side, this model is fast and cheap, and it's fantastic for throwing together great looking demos at a moment's notice.
Lightweight Dashboard Model
Here’s how access works in the lightweight model:
1. User visits their dashboard URL, an HTML file hosted
in Windows Azure Blob storage.
2. The dashboard client HTML5/CSS/JavaScript starts
running in the user’s browser, which could be on a PC, tablet, or phone.
3. The web client issues asynchronous Ajax queries to
the dashboard web services to retrieve content. The content is in the form of
XML or JSON data files stored in blob storage. As the data is returned back to
the web client it is rendered as content panels in the dashboard.
Deluxe Edition
For deluxe implementations, you will want domain identity or
web identity authentication of your users and deeper integration to your
systems. For that we will run on Windows Azure Compute and integrate more
formally with internal, partner, or online data sources which might include
your CRM, data warehouse, and other sources of business intelligence accessed via
feeds or web services. The data can be pulled on-demand from its source systems
in real time, or the data can be periodically pushed up to blob storage as in the
lightweight model. For this level of dashboard you will want to combine the
dashboard core with a consulting engagement to set up the authentication and integration.
Deluxe Dashboard Model
Here’s how access works in the deluxe model:
1. User visits their dashboard URL, an ASP.NET MVC3
web server hosted in Windows Azure Compute.
2. User is sent to the designated identity provider
to sign in. This could be ADFS, a web identity, or the Windows Azure Access
Control Service.
3. Upon successful sign-on, the user’s browser is
redirected back to the dashboard web server and the dashboard client
HTMl5/CSS/JavaScript starts running in the user’s browser, which might be on a
PC, tablet, or phone.
4. The web client issues asynchronous Ajax queries to
the dashboard web services to retrieve content. That content can come from
querying enterprise systems, feeds, or web services directly or by returning
recent data snapshots out of blob storage. With either approach, XML or JSON
data is returned back to the web client and rendered as content panels in the
dashboard.
The first prototype
implements some of the above design goals in the lightweight model, all hosted
out of Windows Azure blob storage. More to come!
Next Part 2: HTML5 Video