The cloud computing revolution continues to go full steam, the problem, of course, being the overabundance of subtle “cloud” puns that come from every analytical article covering the subject. For example, “go full steam.” “Steam clouds.” Get it?
Gah, never mind.
Anyway, while it’s true that the simplicity and lower cost of many cloud computing apps compared to the traditional in-house applications may be drawing people to the cloud, very few companies, comparatively, are taking a good, hard look at the performance of the cloud apps.
According to Information Week, 62% of all performance problems with cloud computing are network related. IW surveyed 287 business technology professionals who use, plan to use, or are considering cloud computing services, and asked them what the “biggest bottleneck for performance of cloud-based apps” is. 26% responded with their own bandwidth or connectivity, 21% with overall Internet traffic, and 15% with the cloud vendor’s Internet connectivity. 10% didn’t know, so the actual number might be anywhere from 62-72%, though where in that number is anyone’s guess.
And that’s a problem – only 20% have actually taken the time out to measure the impact of cloud performance, so quantification of cloud app performance becomes guesswork.
According to InformationWeek:
Respondents who’ve put monitoring tools in place and done assessments report the highest level of satisfaction with their cloud applications and had the best ratings of performance over the last year. They likely have realistic expectations and have put technologies and bandwidth in place to ensure service levels. Those who haven’t done any of that? They had the lowest ratings. If you don’t flesh out your data plan and establish a way to respond to issues, you’re setting yourself up for a fall.
Another problem with not monitoring cloud apps is that cloud vendors can’t make a guarantee at anything like we’re used to with SLA; heck, most won’t guarantee that you’ll be able to get your data back, or that it won’t be mutilated by rabid wolverines. And without monitoring your internal traffic, it’s all too easy for a poorly performing cloud vendor to say that every problem is originating from your end – even if the problem is actually originating from theirs.
Essentially, you should treat adding a cloud app just like making any other major change to your network; create a baseline, understand what normal performance is, and monitor the changes.
No comments yet.