Businesses that consider moving their analytics to the cloud should be more concerned with data migration and network connectivity challenges than worries around security, according to Marc Clark, head of cloud strategy at data warehouse vendor Teradata.
Speaking at Teradata Universe in Nice this week, Clark discussed the findings of a customer survey which polled 900 senior business leaders from nine countries. This found that, unsurprisingly, growing amounts of business-critical data will reside in the cloud by 2019. There are clear reasons for this – for the most part the cloud is well suited to cost effectively running analytics on large volumes of data.
However, the survey also found that security remains a concern for those already using the cloud and a stumbling block for those who aren't. Forty percent of respondents see security as a risk, while 25 percent believe hosting data in the cloud will lead to more security breaches.
"It has been consistent for the last decade," Clark said of ongoing security threats. "Of the folks who are concerned with moving to the cloud, their main reason for not moving is security."
This is despite cloud security being "more robust and more reliable than on-premise security".
Clark highlighted a number of high profile breaches in the US, though there are numerous examples in any country. "Target, Home Depot et cetera, they all happened on premise," he said. "None of those happened in the cloud. So why are all these IT departments and all these people concerned about moving to the cloud?"
He said that there is a misconception about the safety of cloud computing: "When you look at Amazon and look at their security certifications, they have a 600-person security department. I don't think the biggest company in the world has that many security people for cyber security."
The second-biggest reason cited by survey respondents was cultural – specifically, buy-in from executives.
However, Clark suggested that there are other hurdles to cloud usage that are not discussed as often.
"The biggest stumbling block between a successful and unsuccessful cloud migration over time is network connectivity and the reliability of cloud," he said. "And that wasn't even mentioned [in the findings]."
Latency, for example, might not seem to be an issue until data is moved from an on-premise data centre into a more remote cloud facility. Clark cited one US Teradata customer which moved its "entire production" to its cloud facility in Nevada, with applications kept on-site in another state.
"They didn't realise they had latency problems until they did," he explained. "This is a big area that people aren't thinking about."
Another area is the sheer scale of the task of moving large data sets from data centres to the public cloud providers. Teradata is both a customer and a competitor of Amazon Web Services, offering analytics on the AWS cloud.
Clark explained: "How do you move a petabyte of data to the cloud? You don't. Because a petabyte of data is hard to move to the cloud. You may have seen Amazon Snowball, which has 80 terabytes, and they'll send it out to you. But that wasn't enough so Amazon introduced Snowmobile.
"No one has found a way to overcome physics, not even Amazon. So this is one of the reasons that analytics is one of the last things to move to cloud – because who the heck wants to move this data all over the place?"
Clark said that the best approach to moving to the cloud differs greatly for customers, and private or managed clouds can be more appropriate in some cases.
"One thing we find is that there isn't one answer," he said. "A lot of times people come to me and ask what cloud I should move to, and I say I don't know, maybe you should move to multiple clouds. There are some that don't think that is a good strategy, that you should have just one cloud, but you have one point of failure. They are all at different places in their journey to the cloud."