Microsoft announced a new version of their cloud, called Project Daytona. I read this Information Week article on it. Project Daytona will allow users to deploy Google’s MapReduce software within the cloud to run some serious, statistics-based algorithms on serious amounts of capacity. I love the idea and I think it shows the future of the cloud. I hate the iWeek article describing it because I think it leads to some of the confusion around cloud.
The way in which the solution represents the future of cloud is that it deploys a highly specialized solution for customers. Today most companies use the cloud for three purposes; reducing costs by takin advantage of economies of scale, shifting the expense of technology from capital costs (datacenters and servers that must be purchased and depreciated) to operational costs (paying Amazon’s monthly bill) and reducing risk (when an idea goes bad it’s a whole lot easier to cancel or reduce an Amazon EC2 account then it is to return equipment to IBM). For most companies, these three advantages to cloud computing outweigh the technological downsides (lower security, higher latency, etc…). That’s today’s reality, but not the future. In the future it won’t be enough to reduce/operationalize costs and reduce risks, customers will expect cloud providers to make their IT solutions run BETTER than they could in the datacenter. This will be achieved not by provisioning simple VMs to customers, but by creating PaaS solutions that leverage platforms/software that the customers don’t have the time or money to put in the datacenter. This release from Microsoft is an example of that. Project Daytona will enable doctors and scholars access to a resource platform that their institutions simply couldn’t afford to implement on their own.
The way in which the article sucks is that it calls out “Big Data as a Service” and later calls out “Data Analytics as a Service”. Project Daytona is a platform, provided as a service (PaaS), that supplies special technologies to aid massively scaling data analytics (aka Big Data). As I mentioned, I think there will be many many many more platforms like this coming out that help provide fit-for-purpose solutions for workloads, if we give them each their own “aaS” name the world will become far too confusing. How about we just agree: