SaaS Reliability

Blogs, Twitter and the media go wild every time a major provide of web (or cloud) services -- we can reasonably label all of these as SaaS -- has an outage. The most recent object of attention is Salesforce, the top provider of SaaS CRM (customer relationship management). Others in the news in the past months have included: Google's Gmail, Microsoft Bing, RIM email, Twitter, among many others.

These failures are not too surprising since no matter how well you construct your service, failures can and will occur. You can make everything redundant, rigourously test new hardware and software variants, geographically-distribute hosting sites, hire the best people in the business and implement catastrophe plans that anticipate the most unlikely events, both natural and man-made. When all is said and done, what you have accomplished is a reduction of the probability of an outage, or the duration of an outage when it does occur; however, reducing the probability to zero is impossible.

Outages of one type or another have been going on for as long as we have been using technology. It is simply unavoidable. The reason these recent cases are prominent is because each service has many millions of users. Therefore, when an outage occurs, large segments of the population are impacted at the very same time and they talk to each other about it. When the outage affects only one company or a small number of people, you are unlikely to hear about it unless you are directly affected. After all, what do you care if there was a power failure in, say, Owen Sound, unless, of course, you happen to live or work there?

There are voices of reason among the clamouring hordes, although you may have to dig to find them. Here is one who I think gets it right:

"...they shouldn’t distract us from the fact that most corporations and private institutions are equally, if not more susceptible to similar operational problems.

The only difference is that the IT staff within corporations and private institutions only have to support and report to a single customer,..."

Not only that, other than medium to large organizations, companies generally don't have the budget to build a top-notch IT department and network that has all the knowledge and attributes needed. Quite sensibly, more and more companies are moving to SaaS for the same reason they moved to outsourcing in the past: they shift non-core support or other specialized tasks to companies that are focused on those areas. They are then free to focus on their core business. It's efficient and effective, and everyone wins -- the company, its customers and their suppliers -- when it's done right. If it's done wrong they will lose business and will therefore have an incentive to try again.

Despite the Chicken Littles who cry that the sky is falling whenever one of these large outage events occurs, we are better off with SaaS and outsourcing. Hopefully the media will eventually tire of this silly scare game and move on to another irrational obsession.