Virtualization Technology News and Information
Data Protection in the Era of Hyperconverged

Written by Goran Garevski, Vice President Engineering, Comtrade Software

It's fascinating how IT has evolved from its beginning. Initially computers started as basic batch processors, albeit very large. Then as if they weren't large enough to begin with, they were made bigger, and bigger, and bigger...Do mainframes sound familiar? That is, until the concept and cost of IT at that time exploded into micro components, resulting in a tectonic shift toward open systems IT.

It didn't take long before everything started to scale-up again. Starting with siloed servers with direct attached storage (DAS), then siloed servers with virtualized storage, then virtual computing with virtualized storage. IT organizations ended up with a heterogeneous infrastructure with questionable utilization, maintenance nightmares and scalability issues.

Enter the era of virtualization and cloud computing and the rise of hyperconverged infrastructures. Hyperconverged systems addressed the major challenges of open systems by:

  • Simplifying usage and maintenance
  • Enabling high resource utilization
  • Providing scale-out capabilities

Evolution of Data Protection

Data protection was also evolving throughout these shifts and as new concepts emerged. In the mainframe world, one expected data protection to be integrated in the system. In the open systems environments, data protection was also "open." Almost too open.

Anyone familiar with agents, media servers, backup servers? Let me help you. What about dedicated backup admins, backup implementation projects measured in months, professional services for upgrades, PhD pre-requisites for finding your way through the backup console? Did I mention upgrades were also measured in months?

But again, it was not perceived as a problem. As it was a similar approach for everything, not just for data protection.

Virtualization as Disruption

Although, not a new concept, server virtualization introduced major disruptions in IT: new approach, new workflows, new I/O patterns. Also, by focusing on specific problems like VM backup and recovery, and empowering the virtualization admins, some new VM data protection vendors turned the world upside down bypassing the enterprise backup tools, and conquering the virtual on premises IT world. Flexibility, integration and concept alignment. Sounds familiar?

The downside of VM-focused data protection was handicapped application data protection. Moving blocks left and right is not enough, one needs to take care that the business data is consistent when backed up and recoverable when needed. But it was fascinating how the industry closed its eyes and continued to explore a VM-centric movement.

Hyperconverged Data Protection

New IT architectures are attractive as in many situations they are used within products to address specific use cases. The most definitely holds true for hyperconverged architectures. New data protection vendors did emerge with internal hyperconverged architectures, but focused rather on legacy challenges of dispersed backup infrastructure rather than being ideal backup solution for the new on-premises clouds (hyperconverged).

The truth is hyperconverged data protection vendors have major technology, feature and cost overlaps with the hyperconverged infrastructure vendors. Which begs the question, would you buy a new house just because you need a garage?

Data Protection for Hyperconverged

Learning from data protection history we see that the most successful data protection solutions in each era were the solutions that were focused, conceptually aligned and INTEGRATED with the respective infrastructure. The key word here being, "INTEGRATED."

Hyperconverged infrastructure is not just an infrastructure. It is a completely automated cloud, that scales with the business needs, and is simple to use and provision. It comes with an advanced management stack, self-service portals and orchestration capabilities. In many ways, it is like the public cloud, just on premises. It's your LEGO IT - you can use it, play with it and hug it. But, wait a minute! How do you protect it?

When you try to protect HCI with enterprise backup tools (the ones from the open systems world) it just doesn't feel OK. It costs even more. Imagine you build your HCI in a few hours. Great! Then you implement the backup in a week, you have zero integration in the HCI workflows, and it usually requires a separate IT group to sync with. And then you need to upgrade it... That means budgeting, planning, time, professional services. Repeat that again for the next upgrade of your backup solution or of the HCI solution. The overall experience? Like buying a new car without a stereo and having the whole orchestra (read professional services) following you, each one with its own instrument (management tool) and on different transport vehicle (devices).


What about the virtual backup tools? Can they do the job? Of course they can, but let's see how. People that were around the software industry know that once you burn your concepts in the initial architecture of the product it is almost impossible to change them as new concepts arrive. Virtual backup tools treat the hypervisor as a first class citizen, providing slightly increased usability than the enterprise (open systems) backup tools. Meaning they pretty much ignore the rest of the management and technology stack. So all of the disruptive and valuable concepts of the HCI are ignored. Typical symptoms: VM stuns, multiple management consoles, limited automation. Like buying a brand new car with a huger speaker attached on the roof.


To really work, the HCI concept requires fully integrated data protection. Point. And, it is not just the ability to integrate into HCI's snapshotting API. Any storage vendor provides freemium software for that. If the infrastructure is simple and invisible in the HCI world, the same should apply for the data protection. It should have native integration into the overall HCI workflows, self-service portals, and application life-cycle management portals. It should be a service. Ideally, Data Protection as a Service (DPaaS). Like having a brand-new car with premium sound system that feels a part of the car. Like it should be.


Selecting Data Protection for HCI

At the end, it is a decision for the everyday admin and IT manager what should be used to protect modern HCI environments. Each IT group has its own budgetary and operational specifics. However, there are three things to be careful about:

  • If your backup solution cannot keep it up with your HCI infrastructure something is wrong
  • If your backup solution costs almost the same as your HCI infrastructure something is very wrong
  • If your backup solution is not manageable by your HCI stuff something is really, really wrong.

At the end, what's the experience with your sound system?


About the Author 

Goran Garevski 

Goran Garevski, Vice President, Engineering, Comtrade Software

Goran Garevski joined Comtrade Group in 1993, reaching the highest technical level in 2001. From 2002 to 2004, he was CTO of StorScape (storage resource management startup). Following that he has held various business development and management roles within Comtrade, including his current role as VP of Engineering for Comtrade Software.

Mr. Garevski is a storage and data management industry executive with a unique combination of deep market understanding and comprehensive technology insight. Over the last 20 years, he has led and contributed to numerous breakthrough projects that have influenced the storage industry. Mr. Garevski has extensive experience in data protection, virtualization and cloud storage. He holds a BSc in Computer Science from the University of Ljubljana. 

Published Thursday, October 05, 2017 7:35 AM by David Marshall
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<October 2017>