The average data center creates 8-10 copies of application data for a variety of purposes ranging from data protection to development to analytics reporting. The process to create these copies is inefficient and the cost to store them is expensive. These problems have led to a new breed of solutions that do more than just protect data; they manage copy creation for the enterprise. Delphix recently entered this market with its Copy Data Engine.
Copy Data Management Basics
The goal of a copy data management solution is to reduce the inefficiency of making and storing dedicated copies for each application. This is typically done by creating a single, second copy and then creating writeable virtual copies to be used by other processes like development, testing, reporting and analytics. The second copy is typically stored in the native format, instead of a proprietary backup format, so that the data can be interacted with directly by these other processes. A third copy can also be made, via replication, at a secondary site for rapid recovery from a data center outage.
Copy Data Management Differentiators
The key differentiator between these solutions is how and where the secondary copy is made. Some copy data products extend an existing storage system’s snapshot capabilities. This approach offloads much of the copy data management overhead from the application’s servers to the storage system, but does require a specific storage system investment.
Other solutions essentially create a virtual storage infrastructure and implement low level write splitting on each host to direct new data to a secondary storage system. This removes the need for a specific storage system for copy data but does place a potential burden on the application host. It’s also a relatively ‘brute force’ method of implementing copy data management since every I/O is copied, regardless of the value of that transaction to the environment.
The Delphix Difference
The Delphix Copy Data Engine adds a new method to creating the second copy. It runs at the application level. First, like the other solutions, it creates a full copy of the application’s data. But this initial copy is compressed, reducing its size by 50% or greater.
The Delphix Copy Data Engine then records changes to the production application on an ongoing basis via log shipping. As a result, Delphix has a timeline of all of the data and the changes that have occurred in the production application over time. Delphix calls this a “TimeFlow”. From the TimeFlow, Delphix can create point-in-time virtual copies of the production application instances. It can even roll back a database to a previous point in time by applying these changes, like a TiVo can replay a video. Because each copy is initially created from existing and available shared data blocks, copies can be created in minutes, even if the copy is a multi-terabyte database.
Finally Delphix performs these functions on any storage in the data center and across various organization locations. Those locations can also include cloud providers like Amazon.
The application centric approach is very appealing. Instead of copying everything, it provides intelligent copy data management, only copying the critical data. It provides very clean access to data by using the log information that the application creates and it allows the copy data to be hosted on virtually any storage system including the cloud.
In 2015 data protection managers are going to be hearing a lot about copy data management. Its potential to do data protection better as well as to do more than just protect data is very appealing. Delphix deserves strong consideration as data protection managers begin to investigate these solutions.