In-Depth
Composite's Discovery Appliance Brings Self-Serviceability to End Users
What business users really want is self-serviceability, says Composite Software -- and that's just what Composite's new Discovery appliance gives them.
In the data federation space, Composite Software -- launched half a decade ago -- has established itself as a specialist to be reckoned with.
This week, Composite will try to recast itself as a new kind of specialist -- namely, as a force in the data federation appliance segment.
On Monday, Composite launched Discovery, a search appliance that it says fills a gap of sorts: namely, the "chasm" that stretches between traditional BI offerings and newer BI search tools when it comes to ad hoc query support.
What users of ad hoc queries really want is self-serviceability, Composite officials claim; what most BI or BI search tools give them, they say, is anything but self-serviceability. BI tools, for one thing, are too difficult to configure (much less fine-tune) when it comes to ad hoc query usability, contends Bob Eve, vice-president of marketing with Composite.
Search tools, on the other hand, don't have the analytic chops of their BI counterparts. In other words, Eve argues, the existing tooling on either side of the divide isn't getting the job done, so what's needed is a hybrid. That hybrid, Eve says, is Discovery.
"[Ad hoc query analysis] is something that falls in the gap between BI and search. BI takes some time to set up. BI applications themselves are inefficient … [when it comes to] dealing with new things quickly, and they require [the intervention of] IT. Business users can't do it themselves," Eve maintains.
"Search tools, [on the other hand], are very good at finding stuff, but when it comes to structured data, they don't understand very well the fact that customers link to invoices and link to shipments, [for example], so they can see the customer name but are they inferring relationships through that?"
Discovery restores some measure of control to business users, says Eve. The average business manager spends two hours a day looking for information, he argues, citing research from services giant Accenture. Once the information is found, almost half of it turns out to be useless. Narrowing that gap -- increasing the signal-to-noise ratio (or eliminating noise entirely) -- isn't something that, in most cases, the business user herself can do, depending instead on IT to tweak SQL queries, connect more data sources (or accelerate data delivery windows), and so on.
"Business users normally have to depend on IT for access to data for analysis or decision-making," he observes. "They have to wait for IT to connect the [data] sources, approve their queries, install or configure their [client-side] software. With Discovery, they have the ability to do this on their own. They can discover seemingly disconnected things. They can look at the metadata, so maybe they'll start to see [that] because of these primary keys between these tables, because of the values of the data … there is a relationship."
As an appliance, Discovery simply plugs into an existing infrastructure, Eve says, so it's a comparatively easy sell. Composite isn't talking about a hugely disruptive implementation -- with hardware and software purchase orders to approve (and attendant installations to supervise) -- and IT, for the most part, has proven itself receptive to the appliance model.
Intriguingly, however, Composite plans to market Discovery to business users, based largely, Eve argues, on the fact that these consumers are frustrated by what they perceive as a lack of responsiveness on IT's part. "It's definitely targeted at business professionals -- you know, business users. That's a little different for us. We've typically sold to IT, to integration competency centers," he indicates.
Won't this invite discord of a different sort, however? After all, data management (DM) groups are notoriously protective of their finely-tuned, well-managed DM infrastructures. With this in mind, won't IT throw up people or process roadblocks if business users go out-of-band to acquire Discovery?
Not necessarily, says Eve. "People have well-tuned, well-managed infrastructures, but that doesn't mean that they don't have a question that comes up this afternoon, or tomorrow, that -- because they're down on the resource list -- [IT] just can't get to rapidly enough. This leads to discord [between business and IT], so what we've found is that [IT] is very receptive to [the] Discovery [value proposition]."
As for the Discovery interface, Eve says it's largely a point-and-query proposition. "The first-level UI is just a search box, and that has an API for it, so your IT [group] can customize that if you want them to. The same for the second-level display. That's also pretty simple, and that has an API, too," he says.
Beneath the covers, Discovery touts Composite's EII special sauce to connect to and assimilate data from disparate data sources. From there, it integrates structured data on-the-fly into rationalized result sets and then presents its results in a tabular format. Discovery result sets aren't static, either: users can navigate across related objects and use the linkages unearthed by Discovery to drill down into answers.
Elsewhere on the analysis tips, users can filter and refine discovered data, export result sets to Excel, and (in the reuse department) create sharable "recipes" (canned search-and-analysis procedures) and capture "folksonomies." If users identify particularly valuable "recipes," IT can, in turn, codify and instantiate them by creating official reports (with canned connectivity to the requisite virtual or physical data sources), Eve maintains.
"[Discovery] really does close the gap. It lets business professionals handle their own fast-turn, ad hoc reporting requirements without incurring the latency and rigor of IT's own involvement. IT wins because it gets freed up to work on other projects," he concludes.
Composite touts two Discovery pricing models: the first, a perpetual license, costs a flat $150,000; Discovery is also available on a subscription basis, according to Eve, which costs $7,500 for initial setup and $4,000 per month thereafter.