Altosoft Insight: designed for operational BI
Altosoft has just released version 2.1
of its Insight product, which has been designed specifically to
serve the market for operational BI. Since you may not have heard
of them I should say that Altosoft is US-based, founded in 2001.
Development is in St Petersburg and the company also has an office
in the Netherlands. However, before I go on to discuss the product
we need to be clear about we mean (or, at least, what Altosoft
means) by Operational BI.
Operational BI tends to be used rather
vaguely to describe anything that involves the processing of
queries in real-time. However, there are, in fact, two aspects to
this. The first is where you have a call centre, say, which needs
to run a query against the data warehouse because a customer is on
the phone to you. Here the query would be embedded in the call
centre application and run as required. The second type of
requirement is where business events occur that you want to monitor
in real-time, either to drive real-time dashboards or generate
alerts or for business process optimisation. It is this second type
of requirement that Altosoft is addressing. We might refer to the
call centre query as transactional BI.
Now, the most interesting about Altosoft
is that it has been specifically designed just to support these
sorts of Operational BI requirements. Moreover, with the arguable
exception of SeeWhy it is the only product that I am aware of that
has been so designed. Moreover, it does not require a data
warehouse (you can persist information in an Oracle or SQL Server
database) though you can access warehouses (and operational
databases and message queues and so forth) if you need to, and nor
is it based on a complex event processing engine, as is
SeeWhy.
I wont go into the architecture in any
detail because it is relatively complex (though hidden from users)
but it is worth mentioning some notable points. The first is that
the environment is based on defining what the company calls
archetypes, which are high level high level abstractions of
business processes (you can load BPEL directly into the product),
and which are built into archetype flow diagrams. You then map
source events to archetype events and then incrementally define
such things as hierarchies, distribution analysis (for
comparisons with past performance), user privileges and so on until
you reach the point of deployment. All of this is highly
graphical.
At the front-end, building a dashboard in
Insight is really, really easy: it is based on the use of Ajax and
Flash technology (again, hidden) and you simply use a drag and drop
approach, from within your browser, to create your dashboard.
Moreover, because the software understands the data that you are
working with (thanks to the product’s Discovery Server, which sits
between your archetypes and your data sources) it will
automatically prompt you with relevant options, for example to
create relevant dimensions for slicing and dicing.
One particularly cool thing that Insight
does is that, in the latest release, it can generate alerts against
forecasts. In other words, on the basis that it can compare now
with the past, it can extrapolate from today’s events to predict
future trends and thus raise alerts rather than waiting for bad (or
good) things to actually happen.
The long and the short of it is that
Altosoft Insight represents seriously interesting technology: it is
definitely worth a look if you are interested in operational
BI.