How Scale-Out Architecture Fixes Backup

By Greg Page

In his recent blog, “Why Scale-Out Architectures Break Backup,” George Crump said three things that stand out:

  • “Scale-out architectures are the logical solution to the scale-up challenge”
  • “…for the most part, only the backup storage hardware scales-out”
  • “Now there is the reality that organizations expect backup software to evolve and do more than just backup”

George is right. In my previous blog, “What is Scale-out Architecture,” I discussed what a scale-out backup architecture is and how it fundamentally changes the way organizations purchase, deploy, manage and upgrade their backup environment. Now let’s take a deeper look at how scale-out backup does that.

We’re all familiar with the typical IT buying cycle. We plan and size for three to five years, attempt to account for growth, then hope nothing unforeseen happens and we guessed right on the growth. At the end, we’ve either got too much or too little, usually, and that begins an out-of-cycle request for more servers and storage. Only using scale-out storage doesn’t solve this problem.

When it comes time to deploy, multiple IT teams have to coordinate. Storage has to be installed or provisioned, servers installed, backup software installed, configured and tuned.

That’s just the initial net new deployment. If the deployment is adding onto an existing environment, sizing has to be done to see if it’s capacity, compute, or network that is needed or

some combination. Then we have to figure out if there is enough capacity on existing storage or if more needs to be purchased, installed and provisioned. 

Another issue that organizations often run across is the upgrade and refresh cycle. When gear is off-lease, deprecated or up for renewal, we often have servers and storage on different lease or maintenance cycles that have to be upgraded independently. Along with that are the forklift upgrades necessary to replace storage arrays or tape units. There are often lengthy migrations of data, cutover and outage events that must be planned and executed. This is definitely not an ideal situation.

Properly designed scale-out backup solutions, such as Commvault HyperScale™ Appliance or HyperScale™ Software, solves these issues. Because the software, storage, compute and network are converged in a scale-out cluster, it changes the entire life-cycle of the data management infrastructure (more on why I switched from “backup” to “data management” in a bit). Planning for three to five years is a thing of the past. Planning for six to nine months is far more realistic because expanding the scale-out environment is simply buying a few more nodes, which includes compute and storage. Because organizations can much more easily and cost-effectively expand, they are able to buy only what they need when they need it.

Deployments become a simple matter of racking and connecting the nodes and running an install wizard. It becomes so easy that a customer I worked with recently deployed 1.5 PB of Commvault HyperScale in just over an hour after being shown how to do it only once.

In a scale-up model of deployment, organizations often have mixed generations of servers and storage, deployed separately on varying lease schedules. Timing upgrades, replacement of gear with lease schedules – while trying to meet the needs of the business – are a challenging, ongoing exercise in the scale-up world. Compare this to a scale-out environment where nodes are simply added to the cluster as needed with multi-generation support and the difference is vast. From planning to purchasing to deployment to retirement of compute and storage, the process becomes predictable and simple; two extremely important things in business.

To another of George’s points, “… organizations expect backup software to evolve and do more than just backup,” he’s not wrong. Organizations can no longer afford to have separate backup, archive, deduplication, copy data management, reporting, big data and cloud migration products. In many cases, organizations will have multiple, overlapping backup products to handle databases, their virtual infrastructure and file/OS environments. This leads to a lot of complexity that is difficult to manage. Commvault, on the other hand, is recognized as the most comprehensive solution to bring these use cases into a unified, converged platform that minimizes how often data must be touched and how many copies are needed to perform these functions. It’s much more than just backup; it is a true data management solution.