Today, Microsoft announced a whole slew of enhancements to Microsoft Azure IaaS backup capabilities. While the announcement is focused on native backups (Azure to Azure), there is one adjacent development that really makes me excited: the new feature called Snapshot Extension. This is designed to allow Independent Software Vendors (ISVs) image-level access into the Azure VM backup snapshot data, which in turn opens up pretty much endless capabilities, including backup of Azure VMs back to your on-premises environment. And I am particularly excited for this, because the functionality looks to be implemented in the exact same way that I’ve been asking Microsoft Azure data protection folks to implement it – so I like to think my feedback was considered during the design.
But let me step back a little and explain why performing backups of your VMs running in the public cloud back to on-premises environment is so important. Some of you may ask – why do so, when all public cloud vendors already provide built-in backup service? Well, here is the story I like to tell in response to that.
A couple of years ago, one leading public cloud vendor experienced a huge data loss, which caused around 100,000 of their customers to lose all of their data. And two weeks later, this vendor reported that the data for 98% of affected users had been recovered. So far so good, great success at recovery it seemed. But not so fast, let’s look what these numbers mean from the impacted customers perspective.
- There were as many as 2% – or 2,000 – customers who still did not get their data back even after two weeks. And we cannot know if they all ended up getting their data, or if it was lost forever. The question we all should ask ourselves is whether we want to become one of those 2,000 people one day? Or would we rather ensure this to never happen.
- Moreover, this means that around 50% impacted users, which is 50,000 by the way, did not have their data back for almost a week! Just think about this number. This is really not the kind of availability your business will tolerate in the 21st because honestly, even 30 years old data protection technologies provided much better RTO than one week. And these days, businesses expect availability 24/7.
- Finally, 100% of impacted users had ZERO control over the recovery process. They were basically all at the mercy of the cloud vendor, so not only they did not have any ability to accelerate the recovery process – they could not even provide their businesses any estimated WHEN they will be up again. I am sure that many IT heads rolled during those two weeks…
Luckily, most people who seriously considering running production workloads in the public cloud realize these are road blocking and show stopping issues. As I once noted in my discussion with Microsoft Azure team, I strongly believe that lack of robust backup solutions is one of the biggest inhibitor of public cloud adoption – and apparently, most public cloud vendors are not even realizing this!
Now, of course there are ways today to backup data from VMs running in the public cloud to on-premises using the old-fashioned agent-based backup approach. In fact, we had plenty of our Veeam Endpoint Backup FREE beta testers using the tool to protect their Azure VMs. This can certainly be done, but requires playing around with networking to the Azure VM to get the produced local backups back to your data center, and of course these copies require lots of bandwidth and time to be performed.
But, even bigger issue are the agents themselves. In 2015, few people need to be educated why agent-based backups just do not play well with virtualization in the private cloud, and why host-based backup is the way to go. Aside from agent management nightmares, those thousands of agents kicking in to do their job steal tons of CPU from production workloads, reducing the VM consolidation ratio dramatically and thus increasing Total Cost of Ownership (TCO) of such a solution. But in public cloud the same issue is multiplied by the fact that you will also be paying for all these activities, so TCO of such a backup solution goes through the roof.
Last but not least, customers want availability of VMs no matter where they run – whether in their own data center or in the public cloud. They want to ensure their data is protected in the way that ensures RTPO < 15 min for ALL data and applications. Yes, we have spoiled our customers as much. And they realize that the easiest way to achieve this is to ensure that all their workloads are being backed up into a Veeam backup file, no matter where are they running. Only this way are you able to leverage a wide variety of recovery methods and options provided by Veeam, no matter where and how this backup file was produced. In fact, the same backup file format is one of the features of Veeam Endpoint Backup FREE our beta testers totally loved.
All these reasons are why we keep hearing requests from our customers and partners to support backup of public cloud workloads to on-premises with efficient and scalable host-based backups. The need is definitely there, and Microsoft has just made a huge step in meeting this need. As a Microsoft MVP, I was recently briefed on Snapshot Extension capabilities, and I firmly believe that this technology will revolutionize Azure VM data management. Further, I feel these new capabilities will have a similar impact as the introduction of Changed Block Tracking (CBT) by VMware once had on vSphere backups. And ultimately, this will come back to Microsoft in the form of a huge increase of Azure IaaS customer base! Just think about Veeam customer base along, which we expect to exceed 200,000 companies next year! Bravo, Microsoft – what a smart move yet again to enable your partners to deliver value based on your platform!
Of course, the next question all of you will have is WHEN Veeam will have the solution I am talking about, and what will it be able to do? While as always, we cannot share our roadmap for competitive reasons, you can be sure that we have been waiting for this capability to be introduced for a very long time, and have some exciting capabilities to play along already implemented, or at least prototyped – and even more ideas in our minds.
As I like to say, backup is really easy – it’s other things that make it hard to deliver an availability solution that will meet modern data center needs. And due to their specifics, public clouds present many new unique challenges around backup and recovery that we never had to face in regular data centers. Which means that the potential for innovation here is absolutely immense. And there is nothing more exciting for me than to get out on this huge new field of opportunities along with the most innovative R&D team in the world. This will definitely keep us busy for the next few years ;)