There are many companies selling search software these days. And no doubt you’re evaluating some of our competitors. But the simple truth is that the competition cannot offer the features and performance of a Dieselpoint solution. So let’s look at the specifics:
Some of our competitors say they support navigation using structured document metadata. The truth is, they don’t.
Most of our competitors wrote their core search software long before navigation became an important part of modern search engines, and the software simply wasn’t designed to handle it.
The result is greatly impaired search performance. For example, a Dieselpoint competitor built a proof-of-concept system for a major engineering firm with a database of 5 million XML documents. The execution time for a moderately complex navigational query was 22 seconds. We built the same proof-of-concept system for the client. The execution time for our queries was 200 milliseconds.
For those competitors who do support navigation, full-text search is an afterthought, with poor integration and performance.
Only Dieselpoint was engineered from the ground up to search both unstructured and structured data at the same time, using the same index structure.
Almost nothing is more expensive than a programmer. If your programmers must spend days, weeks, or months learning a new tool, then the time and money required to develop your application increases tremendously.
Dieselpoint Search’s API is clean, straightforward, and logical, which means that your developers can write applications quickly — and cheaply.
Virtually all of Dieselpoint’s customers agree on this one point: the APIs for Fast, Endeca, Autonomy and the others are large, complicated, and very time-consuming to implement. Dieselpoint’s API, by contrast, was very intuitive and could be implemented in hours or a few days.
What’s more, our customers agree that there was nothing the competition could do that Dieselpoint Search could not, and Dieselpoint could do a great deal more. Said one customer: “You have got an answer for everything. I love it.”
Simple doesn’t mean fewer features. Simple simply means a better design.
All of our competitors wrote their software in C, C++, or some variant. Ours is written entirely in Java. If you’re working in a Java environment, you know the advantages: an open, easily integrated API, and true cross-platform portability.
Most competitors have a Java interface to their C libraries. But performance suffers when all calls to their software must be translated through a non-native API layer. There are also numerous threading and memory issues. For this reason, some competitors require that you run their software on a physically separate server, and communicate using XML over a socket. This scheme adds unacceptable overhead to a high-performance application.
For OEM customers with applications written in Java, using Dieselpoint Search allows you to be truly cross-platform; Dieselpoint competitors don’t give you that option. Our customers include users of IBM midrange and mainframe computers, which the competition doesn’t support. By using Dieselpoint, an OEM customer preserves the ability to run in these and other, possibly unknown, environments in the future.
The core Dieselpoint search engine ships as a single, compact application. It can run in limited memory if necessary. Installation is simple and hassle-free. Not so for the competition.
A number of customers are using Dieselpoint Search to power both search on a website and search on a laptop or a CD. A single application serves both purposes.
All we can say is try it. For full-text and navigational queries executed at the same time, we beat the competition easily in both query response time and throughput.
One aspect of performance is the size of the index files. For very large document collections, performance is usually limited by the disk rather than the processor. Well-designed data compression algorithms can reduce the size of the index and greatly speed throughput, because the disk is accessed less frequently. Our index sizes are usually significantly smaller than the competition’s.
When the performance is limited by the processor rather than the disk, Java itself is comparable to and often faster than native C code. For more, see our whitepaper Java and the Performance Question.
Most competitors boast of their ability to return relevant content. But here’s a dirty secret: almost all major search engines use similar algorithms, despite marketing claims to the contrary. And their precision and recall statistics (measures of relevance) are similar.
Only Google is different from the others, and that’s because they take advantage of the links available on the Internet. On a corporate intranet or in a product catalog, however, these links are usually not available and Google’s relevance is no better than the rest.
Dieselpoint gives you finer control over relevance ranking than the competition because of its special treatment of document attributes and metadata. It also implements a Forced Rankings feature, which allows you to override default relevance ranking for specific keywords.