A Cloudy Future: SaaS and Balkanization of the Data Center

As I mentioned in my previous post, I will be exploring infrastructure trends, and in particular, cloud computing. But while cloud computing is getting most of the marketing press, there are two additional phenomena that are capturing as much if not more of the market: computer appliances and SaaS. I have just covered computer appliances in my preceding post and we will cover SaaS here. This will then set the stage for a comprehensive cloud discussion that will yield effective cloud strategies for IT leaders.

While SaaS started back in the 1960s with service bureaus, it was reborn with the advent of the Internet which made it much easier for client companies and their users to easily access the services. Salesforce.com is arguably one of the earliest and best examples of SaaS. It nows serves over 100,000 customers and 1 million subscribers with an entire ecosystem of software services and the ability to build custom extensions and applications above the base CRM and sales force functions.  SaaS has continued to see robust growth rates (up to 20%) and will exceed an estimated $14B in annual revenues per Gartner (out of a total industry of $300 to 400B). Growth can be attributed to several advantages:

  • startup costs are low and as you increase in scale the costs are variable based on your usage
  • you can quickly implement to a solid level of base functionality
  • you do not need a large staff of specialized experts to build and maintain your systems

SaaS is here to stay and will continue to grow smartly given its particular advantages and the typically greater margins for service providers. In fact, many traditional software providers are adding SaaS options to their services to enter into this higher margin area. As a consumer of software, perhaps that is the first tipoff for ‘buyer beware’ for SaaS: the high desire of the software industry to move consumers to SaaS from traditional perpetual license models.

What are the advantages of SaaS versus traditional offerings to a software firm? Instead of a potentially large upfront license payment and then much smaller maintenance payments, a SaaS provider receives ongoing strong payments that grow over time — a much more reliable revenue stream. And because they provide software and infrastructure, the total sale is higher. With enough of a base, the margins can actually be even greater than those for traditional software. And, best of all, it is harder for you customers to leave or switch to an alternate supplier. The end result is greater and more reliable revenue and margins, prompting a higher valuation for the software company (and the great desire to migrate customers to SaaS).

As a consumer of SaaS, these results as well as other service factors and business implications must be kept in mind before entering into the service contract. First, some consideration must be given to business criticality of the software. As a business you in essence have three alternatives means to encode your processes and knowledge: custom software, packaged software, and SaaS services. (There is also broader outsourcing but I omit that here). For peripheral services, such as payroll or finance, it is entirely appropriate to use packaged software or SaaS (e.g., ADP or Workday). Typically, there is no competitive advantage beyond having a competent payroll service that maintains confidentiality of employee data. ADP has been providing this type of service for decades to a plethora of companies. But when you begin to utilize SaaS for more core business functions then you have potential competitive risk for your business. Your rate of innovation and product improvement will be limited to a large degree to the rate of advance of the SaaS provider. Even with packaged software your IT team can apply significant extensions and customizations that enable substantial differentiation. This ability becomes minimized with SaaS. For small and medium sized companies these drawbacks likely do not outweigh the benefits of leveraging a ‘good’ platform with little upfront investment. But for a large company, a SaaS course can minimize your advantages over competition. For an excellent historical example, take the case of when I was at JPMC/BankOne and its use of the service provider First Data for credit card software. While First Data provided adequate capabilities and services, the service eventually turned into an anchor. As JPMC tried to drive a faster cycle of features and capabilities to distance itself from the competition, First Data lagged behind in its cycle time – in part because as a SaaS, it was maintaining a broad industry utility. Even worse, once a new feature was introduced for JPMC (and primarily funded by JPMC) it would be available 6 or 12 months later for the rest of the First Data customers. It was not possible to negotiate better exclusivity conditions nor was it in First Data’s interest to have divergent code bases and capabilities for its customers. After detailed analysis, and consideration of the scale economies mentioned below, the only way to achieve sustainable business advantages was to in-source a modern version of credit card services and then customize and build it out to achieve feature and product supremacy. This is then exactly what JPMC then did. While this may be the extreme example (where you are a top competitor in an industry), it represents the underlying limitations of using a SaaS service (in essence, a function utility) for key business capability. The ability to differentiate from competitors and win in the marketplace will be compromised.

Even services that appear to have minimal business criticality can have such impact. For example, if you decide to leverage SaaS for e-mail and chat capabilities, then how will you handle integration of presence (usually provided by email and chat utilities) into your call center apps or other business applications or deliver next generation collaboration capabilities across your workforce? Such an integration across a service provider, your applications, and multiple data centers now becomes more difficult (especially for performance sensitive applications like call center and CTI) to provide differentiating business functions versus comparable services within your data center that are easily accessible to your core applications.

Second, particularly in this time of greater risk awareness and regulation, consideration must be given to data and security. You must be carefully inspect the data protection and security measures of the SaaS provider. While you would certainly review security practices for any significant supplier, with a SaaS firm it is even more important as they will have your data on their premise. Further, it is possible that there is ‘intermingling’, where other customers of the SaaS data is stored in the same databases and access the same systems. Protections must be in place to prevent viewing or leaking of each company’s data. And there may be additional regulatory requirements regarding where and how data is stored (especially for customer or employee data) that you must ensure the SaaS also meets.

In addition to data protection measures, you should also ensure that you will always be able to access and extract your data. Since your data will be at the SaaS site and in their database format, your firm is at a disadvantage if their is a contract dispute or you wish to transfer services to another supplier or in-house. Thus, it is important to get safeguards as part of the original agreement that provide for such extract. Even better, a daily extract of the data of your data in a usable format back to your data center ensures you have access and control of your data. Or alternately, a data ‘escrow’ provision in the contract can ensure you can access your data.  Perhaps the best advice I have heard on this matter is ‘Don’t get into a SaaS arrangement until you know how you are going to get your data out’.

A third SaaS consideration for IT shops is the potential loss of scale and ‘balkanization’ of their data center due to cumulative SaaS decisions.  In other words, while a handful or even multiple SaaS alternatives may make sense when each is considered singly, the cumulative effect of dozens of such agreements will be to reduce the scalability of your internal team and undermine the IT economics. For example, if you have out-sourced 40 or 50% of your IT capacity to SaaS, then the remaining infrastructure must be substantially resized and will no longer operate at the same economic level as previous. This will then likely cause increased costs for the remaining internal applications. Realize that with this ‘balkanization’, your data and processing will then be executed in and spread out in many different data centers. This results in potential performance and operational issues as you try to manage a largely ‘out-sourced’ shop through many different service agreements. Moreover, as you try to subsequently integrate various components to achieve process or product or customer advantages you will find significant issues as you try to tie together multiple SaaS vendors (who are not interested in working with each other) with your remaining systems. Thus, the cumulative effects of multiple SaaS can be far-reaching beyond just the services being evaluated.

So, on the road to a cloud future, SaaS will be a part of nearly every company’s portfolio. There are a number of advantage to SaaS and on the market there are very good SaaS offerings especially for small to medium sized companies. And a few that I would recommend for large companies (e.g., ServiceNow). But medium and large enterprise should chose their SaaS vendors wisely using the following guidelines:

  • think long and hard before using SaaS for core function, processes, or customer data
  • effectively treat such a service as an out-source and have contractual safeguards on the service levels. Closely review current customer experience (not just their references either) as this is a strong indicator of what you will find. And consider and plan (in the contract) how you will exit the service if it does not work out.
  • ensure you have control of your data, and you know how you will get the data out if all does not go well
  • consider the cumulative effects of SaaS and ensure you do not lose enterprise integration or efficiencies as you consider each offering singly.

By leveraging these guidelines I think you can be successful with SaaS and minimize downside from a poor service. My forecast is that it will be a significant part of most portfolios in the next 5 years. This is particularly true if you are a small or medium sized company and your objective is to quickly follow (or stay with) the market in capabilities.  Large companies though will benefit as well when applied judiciously.

In my next post, I will cover the Cloud futures overall and how best to navigate in the coming years (it will be based on both this post as well as the previous post on appliances).

What changes or guidelines would you apply when looking at SaaS? I look forward to your perspective.

Best, Jim Ditmore