Much has been made over the past several days on the seeming total data loss at Microsoft of T-Mobile Sidekick data. How could Microsoft lose all of this data in the cloud? It seems impossible.
The true story may never be know. But it highlights one key point on cloud / SaaS vendors: they all claim wonderful uptime and operational data. Just like every salesperson somehow was in the "President's Circle" or top 10% of earners. Or how almost everyone that graduates from Harvard somehow does so with "honors".
The real question to ask is - what sort of downtime incidents has your vendor had over the past 3 years? How were they handled? And what was the real world impact?
If you hear from a vendor "we're never had any downtime", run, do not walk, to the exit.
No on-line SaaS/web vendor, including Google, Yahoo, Amazon, Salesforce.com, has ever experienced zero downtime. It's not possible nor is it credible. Servers are not 100% reliable -- including the ones that provide redundancy and horizontal scaleability.
The real learning is how vendors handle it.
At EchoSign, we've learned a lot over the past 4 years. First, we provide 2.5 years of 100% fully transparent up-time data 24x7x265 at trust.echosign.com. It includes scheduled downtime - i.e., we don't hide or ignore maintenance schedules. Why? Because it's too easy to blame real problems on 'scheduled maintenance'. And hide them.
We also discuss and highlight the downtime incidents we've had, in real-time, at trust.echosign.com.
At EchoSign we've had two material downtime incidents in our history, and zero data loss. The downtime incidents, one in 2007, and one in 2008, luckily both occurred after standard business hours and had minimal impact on our users. One was due to a total power loss at both the primary and back-up power centers at our primary data center - a once-in-a-decade event as there are multiple redundant power sources in our SAS 70 data center. But, the longer you are around, the more likely a perfect storm can and will hit. The other was a DNS (routing) issue that impacted a relatively small subset of our users at the end of the business day -- but for over an hour.
Evaluating the true reliability of an online services vendor starts with the product, its status in the ecosystem, its reputation and customer base, its scale.
But when the quality-of-service truly matters -- as it does with electronic signatures, when you're closing deals with your end customers -- ask for the real data, and the real stories.
Comments