Last week saw two notable security and uptime snafus. Firstly, cloud sync and backup startup DropBox left a gaping security hole in their application after a product upgrade last weekend. According to The Register, a bug in DropBox’s authentication system enabled anyone to log into any account and access their content. This was fixed after approximately four hours and could, potentially, have affected hundreds of thousands of accounts.
Secondly, just a week before the release of its long-awaited Office 365 cloud suite, Microsoft suffered an embarrassing three hour outage on its BPOS service, reported Business Insider. BPOS (Business Productivity Online Suite, encompassing SharePoint, Exchange and other tools) is the predecessor to Office 365 and launched in 2007 as a single-tenant (i.e. One customer per server) hosted alternative to traditional on-premise deployments. The BPOS service has suffered multiple failures since its launch with its primary cloud competitor, Google Apps, quick to point out that the service was never designed to be delivered via the cloud. Like Google Apps, Huddle was engineered from the ground up to work in the cloud and scale elastically as usage grew.
Like most true cloud companies, the Huddle team is fanatical about uptime. We understand that the reason many businesses choose the cloud is because they want to have access to their data wherever they are and whenever they want. As service providers, if we can’t provide access to that data then we’re failing. And it’s not just the providers who care about uptime and SLAs (Service Level Agreements) – purchasers can get as hung up on the numbers as we do. While it’s important to establish what your service provider’s SLA is, the question you should be asking is: what uptime has your system had over the last three/six/twelve months? SLAs provide reassurance, but no amount of money will compensate for the damage done to your business should the service go down. Your reputation is at stake and your provider should be willing to reveal their uptime. You also need to query if the uptime statistics include scheduled maintenance. Downtime is downtime, whether it’s scheduled or not. This leads me nicely onto the question of the uptime ‘nines’.
We’re very proud to offer all of our customers a minimum of 99.9% uptime and are currently running at 99.992% over the last 90 days (check out the stats in real-time here). But what do these numbers mean in the real world? You’ll often hear cloud service providers talking about multiple nines when referring to their uptime: 99.9%, 99.99%, 99.999% and so on. Rather than simply assuming that you need as many nines as possible, consider what the business impact of the relative minutes of downtime a week and/or month will be. For example, 99.9% uptime equates to 10.1 minutes downtime a week, 99.99% is 1.01 minutes a week and 99.999% is 6.05 seconds of downtime per week.
Given its complete re-architecture, will Office 365 fix Microsoft’s tarnished reputation for providing reliable cloud applications? Early reviews have been mixed, with many analysts claiming the 28 June launch date is simply too aggressive given the number of outstanding issues reported.