We are very pleased to announce the release of Bime V2.5. It includes a lot of new features, including a major one: new compression schemes that will give you the ability to export more data inside Déjà Vu. It sounds geeky, but it is real game breaker and the good news is it is transparent for you, the user. Bime just runs faster while handling more data.
This is part one of a two part series.
Introducing the Cloud Pack Format: a smart compression scheme for data sharing in the cloud
In a nutshell, Cloud Pack Format is a new way of storing the data in memory and in Déjà Vu. The main benefit is it's much smaller than the previous format of Bime that was already compressed compared to the original data source. The main user benefits are that putting data in Déjà Vu is faster, retrieving data from Déjà Vu is faster and that you can put more data in Déjà Vu. For RDBMS for example, we raised the limit in terms of number of rows from 500 000 to 3 million. It's kind of cool to be able to analyze, share and eventually publish this amount of data online in a very simple and economical way: that is no infrastructure on your side and a cost effective model.
Obviously, the number of rows is only one part of the equation: the attribute cardinality (number of different values) and the number of columns also impact the limit. Anyway, it is almost a 10 fold increase in capacity than with our previous format.
This dashboard contains a dataset based on a 200MB datasets with 700 000 lines and it contains all the book titles sold on Amazon since 1951.
Load it here:
How to I enable Cloud Pack format?
In RDBMS, Excel and CSV connectors you have an option that is called compression. You can choose between Medium (the default), No compression (the previous format) and High.
How to I choose the right compression?
Compressed data need to be decompressed at some point. So there is a tradeoff in terms of query performance when you use the Cloud Pack Format. Medium compression is able to answer most of your queries without having to decompressed. In a lot of cases, it will even be faster than non compressed format. That is why now this is the default and the setting we advise if you don't want to fine tune your connection. High compression compress a lot more the data in memory but slow down heavily complex queries. No compression is the old format that is in most cases the fastest to answer queries.
How does Bime handle any type of data set?
Long story short: Bime has a connector for any size on the data continuum. Obviously, with the Cloud pack, Bime can be your BI solution (except may be an ETL complex data manipulation) for small to medium datasets. The more the you go up in chain the more Bime is used as a "client" and delegates the heavy work.
Hint for a future release: the Hive connector.
More data visualization goodness: the funnel chart
Funnel charts are a type of chart, often used to represent stages in a sales process or conversion path in web analytics. This type of chart can also be useful in identifying potential problem areas in an organization’s sales processes.
More data visualization goodness: dual axes
Bar, Column, Line and Area charts now support dual axes. So you can now display two quantitative scales on a single axis (either X or Y).
As Stephen Few pointed out:
A graph should only include a dual-scaled axis [...] when needed to compare data sets that have different units of measure
Dual axis is a great addition to all you can use in Bime to analyse several measures at the same time regardless of their units. In previous versions of Bime you had color and size encoding and Log Axis. Now with the dual axis it's even richer.
More Calculation goodness: Week numbers function
Week number is a tricky thing: do they begin on Monday, Tuesday etc...? This is the main reason why weeks are not part of the standard time dimension Bime build when it encounters a date. Nevertheless, the calculation engine has now an handy function called WEEK_NUMBER that takes a date in parameter and an offset. This offset is the day of week the week "starts" on for your locale - it can be from 0 to 6. If dowOffset is 1 (Monday), the week returned is the ISO 8601 week number.
Better data filters in dashboards
The previous version data filter prompt was not good:
Too much space between them, the text was centered, no max size in width gave scrollbar, selected elements were hard to relate to the attribute you were filtering...
Public facing URL can now be changed
The user can now change the publication url. So instead of the default:
you can have:
This opens the door to the following workflow for managing the dashboard life cycle:
1- Create a dashboard and publish it in the wild by customizing the url.
2- Requirements change. Copy the dashboard and eventually also copy the underlying connections (in the dashboard menu click on duplicate and edit the connections) so everything is decoupled from the production dashboard and you safely change everything.
3- Once your changes are done just change the published url to the production one. It's done. You have replaced your production dashboard. You can eventually delete the old production one if you want and cleanup the old connections.
This simple workflow can be extended to include staging or review steps.
Don't forget to sign up for the free trial ASAP so you can test out the new features.