I have been an interested reader of the Nucleus Research's studies since they launched. For the most part their analysis is insightful and balanced but they really missed the mark in their "Hosted versus on-demand" study.
http://nucleusresearch.com/research/notes-and-reports/hosted-versus-on-demand/
If you have ready any of my other posts on Software as a Service (SaaS), you will realize that I consider SaaS to be a strong business model and a great deployment option, but I do not consider it to be the best model or the death of SW as we know it. It will be a delivery model, but not the only delivery model available. Yes, this runs contrary to a lot of pundits (like Nucleus Research), but let's face it - there are a variety of attributes that make a product successful in a SaaS model, while there are other attributes that would make it less successful. For example, highly integrate applications are not well suited to SaaS given volumes, latency, etc. On the other hand, services with additional value-add components like GSX / Celarix or Constant Contact are perfectly suited. Also, hell will freeze over before government entities and a lot of large organizations use SaaS as the only delivery model for their applications. There are just too many compliance, legal and security issues that they would have to overlook or ignore to make it feasible. For many organizations, economics is also not a good reason - the US Government and Fortune 100 companies each have significant enough data center expertise and economies of scale that very few, if any, SaaS providers can compare to.
Most importantly, just providing a solution in a "multi-tenant architecture" is not enough justification for a SaaS model. Multi-tenant architectures work well, that is how we deployed Celarix, but they also have downsides (which Nucleus forgot to mention). Everybody has to get upgraded at the same time, regardless of organizational change management impacts or interfacing issues. If the application is down or performing slowly, everybody is impacted. I have experienced these issues at Celarix (fortunately very seldom). At IQ, our applications are provided either on-site or on-demand (ASP / SaaS). Each customer gets their own virtual instance of their application run from a cluster of blade servers. The cost economics work well AND the customers can define their own backup, interfacing and security requirements. The choice of deployment option is up to our customers (which is where it should be).
At IQ we use a variety of SaaS solutions such as ADP, ConstantContact and GoToMeeting amongst others. Each of these solutions deliver financial value - I don't need to get servers, communication lines setup or IT resources involved. Each of these solutions also deliver unique business value (e.g. ConstantContact takes care of message delivery and SPAM compliance), but that doesn't mean all applications deliver additional benefit from a SaaS model. That I believe is the fundamental issue I have with Nucleus' report - software that delivers unique value as a SaaS solution should be delivered as one. Otherwise give the customer a choice.
Subscribe to:
Post Comments (Atom)
Posts
No comments:
Post a Comment