Business intelligence (BI) projects still have extremely high failure rates and tie up countless amounts of money and human capital; however, with the emergence of cloud computing and Software as a Service (SaaS), BI solutions no longer have to be unmanageable, risky and prohibitively expensive affairs. Historically, small and medium-sized businesses have not been able to afford the high cost and complexity of BI, while larger business entities became stuck with clunky, brittle and unwieldy BI platforms and architectures that never delivered their promised ROI.
Aside from the steep cost of core BI software (including application development) and its requisite hardware, most of the accepted plans of attack for BI success remain fundamentally stack-based, with different stratum linked together to provide various functions vital to the lifecycle of corporate performance management and strategic knowledge: ETL applications existing in environments squarely outside the data architecture; OLAP and reporting servers performing their magic on separate server clusters; analytical dashboards and front-end applications pulling data from poorly documented federated sources—you get the picture. Not only does such a conglomeration of products necessitate a wide range of expert-level knowledge to extract the maximum performance from each layer, an enormous amount of effort and expense must be put forth to get all the pieces of the BI puzzle effectively working together tandem. A poorly implemented server configuration or oversight in design of one logical component has the possibility to corrupt and compromise the entire BI architecture—resulting in skewed silos of corporate knowledge, ensuring that the alignment between the business and IT remains perpetually elusive. The risk of failure to any BI or performance dashboard project is always high.
Given the current state of BI, it is no wonder that SaaS is gaining more market share of the $18 billion dollar per year BI industry. It has been predicted (by Gartner Group and other IT research firms) that in the next five years as much as 25 percent of new business software will be SaaS releases, requiring little in the way of local installation. With the future growth of SaaS assured, all major BI vendors have been ramping up their offerings in the SaaS space, with an ever-increasing breadth and sophistication.
Although various models of SaaS exist, the most popular varieties greatly resembles the ASP (application service provider) model where the customer owns/leases software on a predefined basis (such as a per-seat, per-month, or per-bandwidth arrangement.) Since BI SaaS applications are hosted by a service provider, SaaS clients generally do not have to fret about the installation and customization of BI software and be sidetracked by the seemingly never-ending chore of keeping software licenses up to date and synchronized throughout the enterprise. A more important reason for taking advantage of SaaS is that it can massively reduce the need for buying and maintaining expensive hardware, especially with respect to data and application servers. Furthermore, the overhead associated with data security can be offloaded to the service provider. Nevertheless, there will be a whole set of concerns to grapple with when a service provider is hosting sensitive and valuable data.
Many companies are leery, and rightly so, of having their most valuable corporate asset moved outside of their enterprise data servers and firewalls and into a cloud where they have reduced control and transparency. However, any SaaS provider worth its salt will be utilizing a robust data center and have an extensive means of application and data security coupled with and impenetrable network topology. Realistically, a provider’s data centers will likely be more secure and fault tolerant than those of their clients.
While due diligence into a provider’s topology is still critical, many times the initial fears about data leaving corporate boundaries will be overblown. To this point, what large organizations don’t already outsource their data (or processes that use their data) to various third parties such as payroll companies? SaaS BI usually means that data will be pulled by (or pushed to) the service provider from a data warehouse or data mart residing on their client’s hardware, so customers of SaaS solutions must pay attention to the potential gaps in their own network and data security setups, in addition to thoroughly reviewing that of their providers. (In most cases, your organization will have the option of retaining its data in-house while leveraging various remote services and BI tool functionality, as opposed to completely entrusting a large amount of data to the provider.)
While transferring huge amounts of data from a corporate data center to a cloud creates a multitude of security concerns, there are sometimes bigger issues. Firstly, when terabytes of data move across even the fastest of internet connections or dedicated network pipes, data latency may occur. Secondly, bandwidth usage fees can escalate, often unexpectedly, to a very ugly number. Data exchange methodologies and costs, both to and from the service provider, must be well understood and negotiated. Another area where SaaS customers often stumble is in the failure to properly integrate SaaS BI data back into their internal systems, making sure that they avoid redundant, siloed and undocumented analytical knowledge.
There is much detective work to be done before becoming a customer and consumer of cloud-based business intelligence. A number of questions must be resolved and facts uncovered before embarking on strategic corporate intelligence missions that are enabled by SaaS. Among the considerations:
- SaaS providers for BI should offer products that have a high degree of built-in templates. For instance, a library of standardized (yet customizable) P&L reports with attached performance metrics and KPIs will be very attractive to BI customers looking for cloud-based solutions.
- Stress testing of the transfer of large data sets to and from the service provider is a must before going live. By nature, most business intelligence platforms wrestle with latency issues due to the need for the integration of data from multiple business segments and systems. Additional latency associated with the movement of data could quickly make things unattractive to users.
- Make sure that the service provider has certifiable expertise in BI and offers consulting services on any tools or functions you will be using. They should be able to jump in and help anywhere, from report design to the creation of data cubes and far beyond.
- Don’t be afraid to be ambitious. Once you have achieved success in going live with one business segment or product line, you may want to tackle several more business or functional areas in the next subsequent release because the services provider’s infrastructure (in particular their ability to quickly scale up so as to crunch large sets of data) will usually be much greater than what is available to you in-house. For example, SaaS customers can increase storage capabilities and processing capability in a dynamic fashion. Demand matching can be easily handled by the provider by adding servers to existing virtual clusters in an automated fashion that is completely transparent to clients and has no negative impact on performance or service-level agreements.
- The service provider should be able to offer you customized historical snapshoting and storage of analytic cubes and query results. Also, actual queries and models themselves should be archived and saved at the host location. Data snapshots and cube archiving can quickly get complicated if proper dating and version control is not implemented, so work out ahead of time what sort of conventions will be employed for archiving “point-in-time” data results.
- Retain a comfort level with your provider’s business continuity and disaster-recovery plans. Are you complacent with the level of fault tolerance that is advertised? In the event of a disaster, is there a way to manually retrieve warm/cold copies of your data from a co-located data center? Do they handle decision support data differently than they handle real-time transactional data when it comes to mirroring and replication?
- Tie your endeavors in SaaS to your Green IT agenda. Chances are that you will be consuming less energy and creating less of an environmental footprint if you outsource your BI functions. Publicize the fact that your organization is able to cut its e-waste and play up on the environmental friendliness of your actions.
- Be careful not to end up with an unmanageable mix of both hosted and on-premise BI systems. Integrity and integration have always been and will always be the biggest barriers establishing the proper foundation for BI.
- Do your homework on IaaS and PaaS, it may wind up creating big dividends for your future BI efforts if you understand the implicit value in these approaches. IaaS stands for Infrastructure as a Service; PaaS represents Platform as a Service. While both these trends are still catching on, they will be considered mainstream solutions for IT managers worldwide in a few short years. IT infrastructure departments will continue to shrink further as cloud computing and data centers become more fertile. A whole new breed of IT infrastructure providers are lining up now to better apply the modular Software as a Service paradigm to their business. Perhaps the best known example of PaaS is Amazon, with their menu of EC2 cloud computing goodies. In short, EC2 lets IT administrators utilize Amazon’s servers on-the-fly, with a pay as you go setup.
- In looking at SaaS BI software or hosted arrangements, the key is to think holistically and understand the entire functional universe of the provider’s offerings – both service/hosted-based products and those that follow a more traditional “site license” model. It is vital that a vendor or provider have product suites that directly complement or drive business intelligence, from data modeling tools to XML-based data integration appliances.
On-demand BI means on-demand licensing; this results in tremendous cost savings for companies that have for years squandered money on software and licenses that filled a niche role or were not as heavily used as anticipated. When the deprecation of hardware assets that supported this software is factored in and amortized over the life of a BI platform, the real costs of ownership for BI projects can be shocking. Every IT director has small projects that, although important to the business, will be in existence for less than a year. For instance a compliance investigation or audit from a regulator will target specific data and require analysis that will not be repeated again; or a non-profit organization may want to conduct a demographic survey over the course of a few months, persisting demographic data for a short time before selling it to a large multi-national corporation. In these cases, on-demand BI makes perfect sense; it is not compulsory or prudent to invest in a data warehousing architecture that will only be heavily engaged for a few months out of the year.
With the advent of SaaS, BI solutions can now materialize in a matter of weeks, at costs far lower than what we usually associate with corporate knowledge factories. Although there will always be a large amount of work to be conducted in the logical design of BI models and cubes (as well as how to best move data and reintegrate it back into corporate analytical systems and dashboards), not being bogged down with large physical data warehouses and RAID devices will provide a level of business agility that was previously only dreamed about. With hardware virtualized, various analytical services outsourced and software consumed on a subscription basis, the realization of “just in time” BI is finally upon us.
No longer will enterprises hungry for BI be excluded from best-of-breed solutions because they do not have the upfront capital needed to implement a BI architecture and hire personnel to run it. The most cutting-edge SaaS clouds which deliver BI applications have the ability to support thoroughly load-balanced, highly available, mirrored multi-tier architectures. BI cloud solutions can now be of the highest maturity levels, resulting in the customizable, scalable, fault tolerant, and secure distribution of the facts, dimensions and measures which propel businesses forward.
About the Author
William Laurent is one of the world's leading experts in information strategy and governance. For 20 years, he has advised numerous businesses and governments on technology strategy, performance management, and best practices�across all market sectors. William currently runs an independent consulting company that bears his name. In addition, he frequently teaches classes, publishes books and magazine articles, and lectures on various technology and business topics worldwide. As Senior Contributing Author for Dashboard Insight, he would enjoy your comments at firstname.lastname@example.org
Copyright 2009 - Dashboard Insight - All Rights Reserved.