Cloud computing offers a lot of promise. By virtualizing hardware and software infrastructure and paying a third party to deliver services as you go on a subscription or usage basis, companies can save a lot of money and time, and speed the deployment of business solutions.
Initially, cloud-based solutions were designed for small- to mid-size companies that didn’t have available IT resources or capital to spend on creating and managing a software and hardware infrastructure. Today, many large companies are investigating the cloud as a way to add new business solutions quickly and augment existing data center capacity. But cloud computing isn’t for everyone, especially in the BI space.
Types of Cloud Offerings
To understand what makes sense to deploy in the cloud, you first have to fathom what the cloud does. In essence, the cloud abstracts underlying services and is a common metaphor for the Internet, which routes data dynamically across a global network based on capacity and other factors. Today’s cloud delivers three levels of services that together comprise a solutions stack: applications, platforms, and infrastructure services. (See figure 1.)
SaaS. Application services are typically called software-as-a-service (SaaS). Salesforce.com, which was founded in 1999 to deliver sales solutions online to small- and mid-sized companies, popularized the notion of SaaS. Salesforce.com now boasts 1.1 million subscribers and has spawned lots of imitators. With SaaS, employees use a browser to access an online application, which is housed and managed by the SaaS provider. There is no hardware to configure or software to install and no licenses to purchase. You simply pay a monthly subscription fee for each user, and you’re up and running.
IaaS and PaaS. In the past several years, the cloud industry has grown, spawning two more services: platform as a service (PaaS) and infrastructure as a service (IaaS). Amazon popularized the latter with the advent of its EC2 cloud computing offering, which lets IT administrators dynamically provision servers in Amazon’s data center and pay according to usage. Many IT administrators now use IaaS as a convenient, low-cost way to maintain development, test, or prototyping environments or to support analytic sandboxes that have a short lifespan. PaaS is the newest addition to the cloud family, allowing developers to build and deploy custom cloud-based applications and solutions. Many PaaS customers are ISVs that want to create cloud-based offerings or enhance them with complementary applications, such as reporting or analysis.
BI in the Cloud
From a BI perspective, all three incarnations of the cloud offer interesting possibilities, but come with constraints. For instance, SaaS offerings are essentially packaged analytic applications. Like their on premises brethren, SaaS offerings need to be highly tailored to an application domain so the solution fits the customer requirements like a glove and doesn’t require endless and unprofitable rounds of customization. And it doesn’t do much good if the SaaS vendor only supports one application out of several because the customer then will end up with a mix of on premise and hosted solutions that are difficult to integrate. So, unless the SaaS vendor supports a broad range of integrated functional applications, it’s hard to justify purchasing any SaaS application.
Data Transfers. Another constraint is that all three types of BI cloud offerings need to transfer data from an internal data center to the cloud. Most BI solutions query a data warehouse or data mart that is continuously loaded from operational systems residing in the company’s data center. Currently, moving large volumes of data on a regular basis to the cloud over the public internet injects latency and complexity into the load process and can become expensive since cloud providers charge fees for data transfers and data storage. In addition, users that query cloud-based data marts using BI tools that run on inhouse servers, then their queries and result sets also travel across the internet, adding more latency and cost.
Given this constraint, BI cloud-based solutions are ideal in the following situations:
- Small companies that don’t have a lot of data.
- Applications that don’t require lots of data, such as development and test environments or small data marts that can be updated quickly.
- Applications in which all source data already exists in the cloud (e.g. Salesforce.com or a start-up company that runs its entire business in the cloud.
- Ad hoc analyses that require one-time import of data from one or more sources. The cloud is proving an ideal way to support data-hungry analysts.
- Report sharing
Data security. Data security is another constraint, but one that is largely illusory. Companies are reluctant to move data outside of the corporate firewall for fear that it might get lost or stolen. Highly publicized data thefts in recent years certainly feed this sentiment. But the fear is largely irrational. Most companies already outsource sensitive data to third party processors, including payroll (e.g. ADP) and customer and sales data (e.g. Salesforce). And when IT administrators examine the data center and application level security supported by cloud vendors, most will say that the data is probably more secure in these data centers than their own! Most new technologies encounter the same criticisms: for example, many thought e-commerce would lead to widespread fraud when it first became available in the late 1990s.
Due Diligence. Nonetheless, companies looking to “outsource” applications, platforms, or infrastructure to the cloud should investigate the cloud providers operations to ensure that they can meet your system level agreements for security, availability, reliability, scalability, and performance. For instance, what is the providers failover and backup procedures? Do they have a disaster recovery plan? Do they comply with SAS 70 data center security protection guidelines?
In addition, you should carefully analyze pricing policies and total cost of ownership. Does the SaaS provider charge set up or cancellation fees? At what point in the future will the total cost of the SaaS solution cost more than if you had purchased a premises-based license?
Finally, you should analyze the vendor’s viability. SaaS vendors take on greater risk than traditional software vendors because their financial model accumulates revenues on a subscription basis rather than upfront. And since SaaS vendors must invest in more hardware and customer support resources, they are prone to suffer from lack of capital. As testimony to the challenge of launching SaaS-based products, LucidEra, one of the first BI for SaaS offerings, closed its doors in June because it couldn’t secure another round of funding. (See my blog: "The Economy Takes Its Toll.")
Conclusion
BI for SaaS offers a lot of promise to reduce costs and speed deployment but only for companies whose requirements are suitable to cloud-based computing. Today, these are companies that have limited or no available IT resources, little capital to spend on building compute-based or software capabilities inhouse, and whose BI applications don’t require significant, continuous transfers of data from source systems to the cloud.
Posted by Wayne Eckerson on June 23, 20090 comments