The reality of (Cloud) Computing

The recent Google Mail outage certainly resulted in a lot of blogs and complaints amongst its users. As part of the large user community, we did feel the impact when our operations were disrupted. Fortunately, this ordeal was resolved two hours later and the overall impact to us in our timezones was minimal, not that I want to downplay the impact to some customers in any way, we all need our email.


It is definitely inexcusable for outages to happen but IMO what happened with Google Mail is unfortunate but, software is built by people and people make mistakes.  We have to be realistic that 100% uptime isn’t that feasible, cloud or no cloud.  We have to understand the risks involved with our data and plan accordingly.   I remember going through contract negotiations in a past life and a customer asking for a guarantee that a particular vendor’s software didn’t have bugs, our legal counsel replied that he couldn’t guarantee it didn’t have bugs, but that he could definitely guarantee it would, and he was only half joking.


In another past life we were using Microsoft Exchange which was outsourced to a third party provider, not only did it cost many more multiples than Google Apps, we also experienced frequent outages and whilst our supplier worked overtime to get us up and running again, they had to rely on support from their vendor to resolve issues, sound familiar?  IMO I’d much rather run with SaaS from a vendor such as Google than an ASP that is relying on a vendor to support them in outages of this kind.


Google has taken some measures with the introduction of Apps Status Dashboard that promotes transparency on their system availability. On top of that, a 15-day SLA credit was given to all its Premier Edition users as compensation.  For us, we have eMail at the host and at our clients, plus we sync our calendars, so that minimised our impact.  I think the major gap is Google Docs, and hopefully, that hole will improve with the release of a better offline tool for Docs, Gears is ok, but not yet complete in my opinion.  Fingers crossed that GDrive is a reality in the not too distant future.  We also use another handy tool, Syncplicity to sync our on-premise data into the cloud, we’ve all had our laptop/desktop fail and lost valuable data in the past, so we use this tool to minimise our exposure with any on-premise data.


As SaaS subscribers, we have researched our vendors well enough that we trust that they have a reliable support, backup and contingency process just as we have a reliable contingency process for our important applications and data.  SaaS doesn’t necessarily mean that the service will not suffer from the occasional hiccups, just like our on-premise applications did before we moved them into the cloud.  As a consumer of SaaS applications, the recent incident does serve as a timely reminder to us to review and identify where we are exposed, measure that risk and if appropriate put a contingency in place.