Data Expedition, Inc.
Articles, events, announcements, and blogs
One of the most common questions I hear is, "how much will it cost me to move my data?" It is a fair question, and crucial to answer before moving ahead with a project. But cloud veterans know that the early answer will always be the same: "It depends." Almost everything in cloud costs money, including metrics you may have never heard of, and the amounts vary by volume, location, and subtle technical details. Understanding how workflows and cloud cost structures interact is crucial to achieving the key cloud benefit of a low-cost service model.In vs Out
The biggest cost influence on cloud workflows is the disparity between uploading data into the cloud, and downloading it back out. In is free, out is expensive. Market leader AWS charges around 9 cents per gigabyte to download data from their cloud, depending on the region and monthly total. This heavily favors workflows that are either archiving data (e.g. backup or archiving to S3 or Glacier) or processing it into smaller forms (e.g. transcoding or analytics). Workflows that involve distributing large amounts of data (upload once, download many times) may find cloud costs prohibitive.
For example, I recently spoke to a customer who needs to distribute terabytes of video files per month between regional offices. The cost to set up direct distribution would be thousands of dollars in up front capital (CapEx), so he was considering cloud as way to shift the cost to a monthly expense (OpEx). But moving that much data through the cloud would cost hundreds of dollars per month in data transfer fees alone, multiplied by the number of destination offices. Over the course of a year, this workflow would cost thousands more to implement in the cloud versus a traditional capital outlay.
But by digging deeper into the workflow, we can find true savings for cloud adoption. In this case, the purpose of moving those raw video files around is collaborative editing. The final product will be tens of gigabytes, not terabytes. Adopting a cloud-hosted, proxy-based, editing workflow would mean that the big data goes into the cloud and stays there. Only the much smaller video proxies and the final product would need to incur internet transfer charges.
Shifting tools, vendors, and established practices is not something to be taken lightly, but by adapting the workflow to the cloud cost model, the desired goal of lower total cost, charged as a service, can be achieved. In addition, eliminating big data transfers during the editing process speeds up collaboration. And, of course, adopting CloudDat acceleration for the initial upload and final download means starting and completing the workflow that much faster.
There are other cost metrics which inform workflow: storage amounts, storage types, I/O operations, computing types, and a vast array of application services. Some costs will amount to pennies a month and can be ignored. Others will balloon to dominate a workflow's budget. Understanding how workflow affects cost, and finding opportunities to save time and money with new workflows, is a critical step to cloud planning. We can help.