I've said before that SaaS providers need to think more about the space between their datacentres and the end user. I summarised this post by saying..
To my mind a good SaaS provider should be interested in this additional real-estate. It represents a space that If MANAGED could be a point of differentiation. A way to get the mass of the adoption curve past their current hurdles and a way to provide businesses (who are putting up mission critical information into the cloud) some sort of SLA (and by association reassurance). It also represents the next evolution of maturity in the SaaS model, SLA’s.
To a SaaS provider, the end users service experience is everything. Service uptime being one of the most critical and basic elements of this experience. Unfortunately for SaaS providers, their end users aren't readily going to discriminate between faults that are within the SaaS providers responsibility or in the Telco's responsibility when making their decisions about the viability of continuing with the SaaS app. To them, the app will just be down.
I'm not alone in getting stuck into organisations who provide services over the web for their lack of up-time. Allan Leinwand at Gigaom is too, only he's found some empirical evidence . (This is the actual service uptime, but as evidence it supports my claims that some of these web applications are too flaky to be considered as business applications)
I acknowledge that these aren't SaaS apps, but the impact can be measured. Look at that Twitter down time, basically one whole working week. Imagine you are the business owner and thats you're business critical app thats down. Imagine not being able to do business for a whole week, imagine paying staff for that week, imagine dealing with customers….
Untill SaaS companies get these type of impacts and address them, to me you are going to have a lot of difficulty taking SaaS mainstream.