SCIENCE

The data generated and stored by enterprises are growing, here are some simple statistics.

• Ninety percent of data volume generated today was produced in the last two years
• By 2020, the Digital Universe will be 44 times as large as it was in 2009.
• 35 zettabytes by 2020. That's 35 trillion gigabytes.

These figures are truly astronomical but they are real considering the amount of content generated by the various devices, digital media, social networking and other sources.

However, what is the usefulness of such vast quantities of data unless users can search them in a manner that will be useful to the information needs of the enterprise.

Hence the Enterprise Search strategy for the large enterprises assumed significance a lot earlier.

Characteristics of Enterprise Search Strategy
Enterprise Search needs are very different from Content search as used by Google and other search engines, where the contents of the website have been crawled, indexed and searched. Here the matching of text strings to the content of a website plays a dominant factor in search results. The following are the ways in which the enterprise search needs differ:

• Enterprise Searches will be meaningful only if they follow a taxonomy or hierarchy in the way the business is organized. For example, any searches about vendors will only make sense if the particular region, business unit and product line that is of interest is taken into consideration. Otherwise the user may see lot of redundant data.
• Enterprise Searches need to consider both structured and unstructured content. While most of the Internet search engines do work on the unstructured data, enterprise searches definitely need to consider a huge volume of data inside the:
• ECM Repositories (Enterprise Content Management)
• Microsoft Office
• Relational Databases
• SAP and Other ERP Applications
• Legacy / Mainframe applications
• Other Unstructured Contents of the new era
• The relevance of enterprise search needs to be highly accurate unlike a regular Internet search since the wrong results may hinder the enterprise productivity. Remember warehouse personnel finding the details of a part based on a search on a vendor name, any wrong result will greatly affect the productivity. The ranking algorithms for enterprise search have to be highly optimized.
• Security is another important aspect of enterprise search strategy. Most application systems within an enterprise are built with authorization and fine-grained access strategy, which means a crawler-based enterprise search approach should not simply bring all the relevant hits, rather it should go by the authorization of the searcher and bring only the relevant details that can be viewed by the searcher.
• More than a full text search: A typical Internet search is all about a full text search of the web pages. However, enterprise search needs to infer lot more information is available in the back-end systems by joining the data between them. For example, is someone is searching about a child part, the user may expect to see the parent part details also. In which case the enterprise search needs to use the taxonomy data stored in the ERP databases and bring the results.

Enterprise Search and Hybrid Cloud
Based on the above, it's evident that Enterprise Search plays an important role in future enterprises. Also the Hybrid Cloud Model provides a perfect platform for hosting an Enterprise Search application, with the following components:

• Enterprise Search Connectors: These consist of several connectors to enterprise content management, ERP and databases; this can be SaaS-based connectivity Services that allow the enterprises to connect to the repositories of their interest.
• Crawl / Parse and Index: This is a highly time-consuming process of extracting meaningful search parameters and taxonomies from the enterprise repositories and indexing them as part of the Enterprise Search metadata. This metadata can be kept on the Cloud platform and the crawl process can be implemented using optimized processes like HPC (High Performance Computing).
• Search Results Display: These are ideally implemented as part of web portals with rich user interfaces and facilitate the searching of enterprise content. These also implement a high level of security in terms of authentication and authorization. These can be web-based elastic web applications that can scale up and scale down depending on the user search needs.
• Legacy Enterprise Repositories (Structured Content): These are the real enterprise data that will be hyperlinked from the metadata available over the Cloud. These data may or may not move to the Cloud based on the organization's needs.
• Enterprise Repositories (Unstructured Content): While this can remain inside the boundaries of the data center, the amount of volume expected on this makes it ideal to be hosted on Cloud Elastic Storage so that the large volumes of the unstructured content is stored and utilized on an on-demand basis.