Release 8.2, Inline Adaptive Deduplication, Maximum Backup, and Suggested Backup
This week Unitrends begins shipping on a general availability basis our Release 8.2 for the Recovery-Series physical appliances, our Unitrends Enterprise Backup (UEB) software, and for Unitrends Cloud. Not coincidentally, we are also changing the way that we enable our buyers and partners to size our appliances. We have eliminated the concept of “Maximum Backup” in our appliance descriptions and removed entirely our “Appliance Navigator” program in favor of a simple table that shows the raw capacity and the “Suggested Backup” of each of our appliances.
So why did we do it? First, a little history. The reason that Unitrends originally had a post-processing deduplication architecture was to increase the performance of backup ingest, instant recovery, archiving, and standard recovery for the most recent backup. We introduced this a few years ago. And for the last few years, we’ve tasked our engineers with something amazingly difficult – build a state-of-the-art inline adaptive deduplication system that continues to have high performance on backup, instant recovery, archiving, and standard recovery. In other words, write software that defies the laws of deduplication – that overcomes the data penalty of the “rehydration” of deduplication.
It took a very long time. It took a lot of early mornings and even more late nights. Stress? Insane stress. But our engineers not only accepted the challenge, they succeeded beyond our wildest dreams.
How much did they succeed? Let’s imagine two companies: Unitrends and a hypothetical company we’ll call “Company B.” Here’s an example of a randomly selected Microsoft Windows 2012 R2 physical server and VMware vSphere virtual machine on two physical backup appliances offering the same amount of raw storage:
But what about maximum backup and suggested backup?
Unitrends has in the past used the concept of “maximum backup” which means essentially the maximum backup we would recommend that would fit on a system with the lowest retention. In our prior post-processing deduplication software, the “maximum backup” depended upon the compression rate of the data being backed up and the change rate associated with that data. Our appliance navigator used an elaborate set of formula that was back-tested against all of our customers to come out with a recommendation that was accurate most of the time. However, with our new adaptive inline deduplication in release 8.2, the calculation gets even more complicated – the size of the maximum backup depends upon the deduplication rate of the data, the compression rate of the data, and the change rate of the data – as well as other factors.
We aren’t soothsayers, seers, psychics, fortune tellers, or any other form of magician. We can’t tell you how many angels will fit on the head of a pin. Of the exabytes of data we protect, we can tell you on average how much backup and retention our customers get within a specified amount of storage. That’s what suggested backup is – a conservative number that should give most of our customers an average of at least 30 days or more of retention. For many data sets, you will be able to backup even more than we suggest with even greater retention. But no matter how much you are able to backup initially, your data will eventually grow beyond the capacity of your backup appliance. This is why we offer Unitrends Cloud. But beyond that, we offer all sorts of other options for retention – from AWS and Google Cloud Platform to rotational disk and tape to fixed NAS and SAN. And of course, this is why we offer Unitrends Pledge – which offers not only evergreen hardware every 3 years but also ensures that the new hardware has 50% more backup capacity than hardware your previously purchased.
It’s a lot like buying a truck to carry a thousand shoes (note: it’s a LOT more complicated than that, but bear with me – it’s a metaphor.) It’s going to be tough to get a salesman to guarantee you that the truck you buy will do it. That salesman doesn’t know what size shoes, what kind of shoes, whether the shoes are in boxes, and so on. And even if the sales person knew all of that today, she doesn’t know what kinds of shoes you’re going to be carrying next week – or the week after – and when you’ll want to carry 2000 or 3000 shoes. So she’s going to help you buy the best truck based on your budget and your needs. If eventually the shoes don’t fit in the truck, she’s going to point to the trailer hitch and help you with a trailer. If you want to carry more shoes without a trailer, she’s going to help you buy a bigger truck – she might even help you buy two trucks to handle your business growth!
Being able to give customers choices is the hallmark of customer obsessed companies. That’s why we offer Unitrends Cloud, archiving, an appliance range from 1TB to almost 200TB, and radically easy to use single pane of glass user interfaces that manage our appliances in a grid. But just as importantly, it’s why we continue to spend so much R&D on features such as release 8.2’s adaptive inline deduplication. We want to be the company that helps its customers recovery more – and that means we have to do more each day for our customers.
