EMC, Avamar, Data Domain, and Unitrends

Recently I’ve been getting a copy of an e-mail that EMC has been sending to buyers who choose Unitrends over the EMC Avamar and Data Domain products.  The first few times I got the e-mail, I shrugged it off.  But after the next few, I thought it would be interesting to discuss this – that it might help in terms of future buyers who are comparing the two products.  So what I’m going to do is to take the body of this e-mail, which apparently has been cut and pasted quite a few times by folks at EMC, and address it on a point-by-point basis.

>>EMC technology provides a more valuable and robust solution [versus Unitrends.]<<

I certainly understand why EMC personnel on would believe this.  I don’t believe it to be true – but then again, I work for Unitrends.  What I’d call out is that quite often, Unitrends and EMC/Avamar/Data Domain are addressing different types of buyers.  What Unitrends does is focus on being the “Apple” of the data protection business.  Our focus is on all-in-one, integrated, easy-to-use physical and virtual appliances, that enable our customers to focus on their business rather than backup.  While our virtual appliance will work with other server and storage platforms (using VMware vSphere or Microsoft Hyper-V hypervisors), even in our virtual appliance we are focused on integrating backup, archiving, public/private cloud disaster recovery, and failover virtualization – so that includes things like physical dissimilar BareMetal, D2D2X archiving, and a lot of other functionality.

>>Our deduplication is more granular, therefore takes less storage and less bandwidth to replicate. (Unitrends openly admits that we have the best de-dedup in the industry).<<

I think the parenthetic aside comes from me.  And I do I believe that Data Domain has one of the best agnostic (non-content-aware)  target-based deduplication devices in the industry.  But what’s missing here is the rest of my statement.  What I tell people is that if you’re looking for the best agnostic target-based deduplication device in the industry, then you have to consider Data Domain.  But what I say next is kind of key – which is that most people are looking for the most backup capacity and retention for the dollar rather than the best niche technology – and that Data Domain and whatever backup software they slap on it combined tends to not only be very expensive, but also have all of the total cost of ownership (TCO) drawbacks of non-integrated backup approaches.

>>We have 65.5% markets share in PBBA, whereas Unitrends is not even mentioned on the report (meaning less than 1%); it’s not even a blip on the radar.<<

This is a reference to the IDC PBBA tracker.  The IDC PBBA tracker tracks both backup appliances (integrated functionality), as well as deduplication devices (doesn’t do backup.)  EMC does have a high market share – driven primarily by a deduplication device (Data Domain.)

Now – contrary to the assertion above – the reason Unitrends doesn’t show up on the PBBA tracker is that we’re a private company and we don’t release revenue numbers to IDC (at least, we haven’t so far.)  But that’s calling out that the good folks at EMC aren’t being factual – it’s more instructive to understand the fundamental fallacy of the argument.

By the logic above, the market share of a backup vendor is by definition a proxy for whether a data protection solution is superior.  Fortunately, that’s not how the market works.  Unitrends grew at almost 90% last year because a whole lot of folks made a decision that we had something that had a lot of value.  Period.  I could note that EMC/Avamar/Data Domain had a much lower rate of growth than Unitrends – but the fact is we *are* smaller than EMC.  Then again, our customers don’t have to get worried about getting lost in the apparently infinitely vast conglomerate that is EMC.

>>We have more people working in support than Unitrends has in their entire company.<<

I think that the most obvious question is – what is it about your products that cause you to require so many support people?

But I think even that (as well as questioning the support expenditure per revenue dollar) is misleading.  Unitrends has a 98% customer support rating as per NPS (Net Promoter Score.)  And that indicates we have some great support people, and that insanely great support for Unitrends is a fundamental differentiator.  But we also have a product engineered from the ground-up to be an integrated appliance.  We have a great deal of intellectual property around that concept – including such things as SNMP trap monitoring to a cloud enabling proactive support.  But at the heart of everything we do is the idea of an all-in-one integrated appliance – and that gives us a decided advantage.  That – and simply the burning passion to delight customers (it’s not coincidental that the title of this blog, which hasn’t changed in the years since I’ve been writing it, is “Customer-obsessed next generation backup, archiving, instant recovery, and disaster recovery”.

>>We provide daily verification checks on ALL the data; Unitrends does not offer anything remotely close to our Data Invulnerability Architecture.<<

We don’t grind on our data all the time – instead we offer not only inline and post-processing verification but also audit modes of failover virtualization (instant recovery) where your software proves itself not theoretically, but by booting and working without customer involvement.  But that’s a little technical – I’d argue that if I were Data Domain I’d grind on the data all the time as well because of the agnostic architecture in which there’s no integration between the backup functionality and the deduplication functionality.  Our very act of creating incremental forever synthetics and our content-aware deduplication validate the data as a side-effect of the backup process – thus rendering not only data integrity but a lower overall load on the system than the cost of not understanding the data backup format and thus needing additional validation/verification steps.

There’s a technical discussion that the letter goes into as well – but given how many words I’ve already written here – I’ll address this at another time in another post.