LESSON - Managed Software-as-a-Service Business Intelligence: A Model that Works
By Larry Hill, Director of Business Development, Full 360, Inc.
The Promise of SaaS BI
In the past few years, there has been excitement in the market about SaaS BI—query, reporting, dashboards, and analytics on a cloud platform with a low monthly cost and unlimited growth potential. Implementation would be quick and easy, with minimal involvement by corporate IT departments. Users would simply provide their data to the vendor and quickly be in full production of reporting and analytics. The SaaS BI vendor would provide technical guidance and support to make the BI project a success. Unfortunately, the broad experience has fallen short of this promise for a number of reasons.
The Task is Rarely Simple
SaaS BI solutions often neglect the underlying issue presented by clients: their data comes from multiple places and needs to be integrated and organized before it is suitable for reporting. Even if much of the organization’s data has been concentrated in a primary file for reporting, there are usually secondary sources of critical business data that must be integrated before comprehensive reporting can be produced. These data sources are usually “live,” in that they generate new detail records daily, so automated feeds must also be created.
So the underlying requirement for many prospective SaaS BI users is for a data mart or data warehouse (DW) to store and organize the data. Low-cost SaaS BI vendors are often a poor platform for data warehousing because they offer standardized data models and minimal customization to keep their costs down.
If a SaaS BI vendor does offer a more customized approach, the cost begins to reach that of the traditional BI and DW models—without the level of control acquired by doing the work in house.
There is no free lunch in business intelligence.
Reinventing the BI Wheel
Many SaaS BI vendors have reinvented the wheel when it comes to BI tools. For “ease of use,” they have recreated tools for ad hoc query, production reporting, advanced analytics, and OLAP, which often provide a facility that only works well in simple scenarios. There are myriad functional nuances in the broad spectrum of reporting and analysis, and few of the new vendors cover this range comprehensively.
The 80/20 rule only works if you are on the 80 percent side of the equation.
Vendor Lock-In
Typically, these new BI tools run only on that particular vendor’s platform. This approach locks the users into that platform because their reports and dashboards won’t run anywhere else. Because the creation of these reports can involve a large investment of time and energy, the user is effectively tied to that vendor, its future path in the market—and its pricing.
Viability is also a concern with new technologies, as seen in the recent demise of two early leaders in the SaaS BI field: LucidEra and Blink Logic.
Platform choice and vendor viability are key criteria in the choice of SaaS BI.
RDBMSs Are Not Enough
These proprietary platforms also lag in terms of performance, typically utilizing an open source database such as MySQL or Postgres to minimize costs. The major RDBMSs are expensive and were really designed for transaction processing, not BI. Without significant investments in computing power, they can bog down in big data analysis and reporting scenarios. It’s not surprising to see that they are rarely used in a SaaS model.
The traditional SQL databases are not up to the task of next-generation BI.
Managed SaaS BI
To take the promise of SaaS BI and make it viable for a broader range of clients, you need to move to the next level of functionality and flexibility, which is a blend of:
- Open source BI and ETL applications
- Advanced analytic databases
- Next-generation systems integration platforms
- Leading cloud computing platforms
Systems integration expertise in both BI and cloud computing is the key to make this best-of-breed, comprehensive BI/DW solution viable for most organizations today.
This managed SaaS BI approach makes possible customer-specific implementations of a robust BI platform and allows for growth in the client’s requirements. It also provides a single source of contact—and problem resolution—for multiple best-of-breed vendors, avoiding the integration and systems management overhead that an internal IT department would face.
Robust Open Source BI and ETL
Open source BI applications are the product of many years of effort by thousands of users in the open source community. They offer the full range of BI functionality—ad hoc query, production reporting, dashboards, advanced analytics, and OLAP—but cost a small fraction of the offerings from traditional vendors. Under a managed SaaS BI model, these applications provide an ideal blend of the traditional vendors’ functionality with the quick implementation of more limited SaaS BI Web sites.
Like its BI counterpart, open source ETL benefits from a broad community of active developers. They have created hundreds of free connectors for a multitude of data sources, and radically lowered the cost of integration with operational systems such as ERP, CRM, and human resources.
Advanced Analytical Database
The platform underlying the open source BI application must be robust, open, and fast, so it should take advantage of recent advances in database technology (e.g., columnar indexing, aggressive compression, shared-nothing architecture, massive parallelism) that the traditional DBMS vendors have yet to embrace—all without sacrificing SQL compatibility.
Leading Cloud Computing Platform
To provide the lowest-cost computing, managed SaaS BI must run on a leading cloud platform. Cloud computing has emerged in recent years as the leader in low-cost, high-flexibility computing. Vendors such as Amazon.com, with broad experience in developing its vast Web site and a focus on being the low-cost vendor in its market, offer their computing facilities to the world at unheard-of prices.
Next-Generation Systems Integration Platform
To manage this powerful service while customizing and scaling it to unique client requirements, the managed SaaS BI platform must utilize next-generation systems integration platforms. These platforms provide the ability to define the infrastructure using easy-to-understand building blocks and open source programming languages. This means that no matter how complicated the client infrastructure gets or how quickly systems architectures advance, the platform can adapt to it. This platform is the “glue” of managed SaaS BI.
It is the synthesis of all of these advanced technologies—open source BI and ETL, revolutionary analytical databases, flexible, low-cost cloud computing power, and nextgeneration integration platforms—that Full 360 has combined into a managed SaaS BI solution that we call elasticBI™.
This article originally appeared in the issue of .