The Doppler Quarterly Spring 2019 - Page 63

And here is our first challenge. There is a metered cost component of moving data between on-premises and public cloud that did not exist in the on-premises-only solu- tion. While most of the major cloud providers do not charge for ingress, all levy some sort of cost for data moving out of their platform. So, while there is no real cost associ- ated with the public cloud reading data from on-premises data stores, there is definitely a cost associated with writing data back to an on-premises system (not to mention the cost of the data that leaves the public cloud platform on its way to the user.) In some ways, this is more of a cost issue than a portability one. Yet the cost of data movement may be so high as to prevent an effective move of the application to the pub- lic cloud. So we make the bold decision to move the data to the cloud as well. All In on the Public Cloud The organization decides it is possible to eliminate the on-premises data store alto- gether (a rarity in most enterprise-grade operations, but a possibility nonetheless). Figure 3: Public-cloud-only application environment We still have a portability issue in that there is a cost (in both money and time) for mov- ing the data stores into the public cloud. While the public cloud vendors have made some tremendous strides in making mass data ingest possible (such as AWS Snowball), the effort associated with moving massive amounts of data between any two locations is still very large. Moreover, while we have made the move to the public cloud, we still have some portabil- ity issues and associated costs to contend with. Expanding Our Reach One of the benefits we expect from public cloud adoption is the ability to move between cloud providers with relative ease. We may want to do this to provide geographic pres- ence, or to prevent vendor lock-in. Technological advances such as containers now allow much more portability of the application code itself than what we had in the past. But we still have that pesky data to deal with, and so we find ourselves in the following situation: SPRING 2019 | THE DOPPLER | 61