The other 90%
There is also the customers side. You have humongous enterprises and very very very small businesses that run their whole IT shop with one, no wait, with half a resource. And then there is the other 90% in between. This is where my point of today is going after.
There are a few blogposts, youtube roundtables and twitter conversations that I read the last few days/weeks that already planted a seed but it was this tweet from Chris Mellor – and more specifically the conversation afterwards – that triggered this post.
Why don’t mainstream array makers have effective answer to Nimble Storage, Tegile & Tintri?
— Chris_Mellor (@Chris_Mellor) August 3, 2013
Everyone involved in the conversations are really respectable people in our industry but most of the answers were dismissive about the innovation in these products at first and secondly on it’s scalability. What about management when you grow beyond first capacity?
Don’t buy today what you won’t need tomorrow
Well, frankly – sometimes – most of the times – you should look for “good enough”. Let me give you a good example here. One of my mantras is “don’t buy today what you won’t need by tomorrow”. With the current speed of evolution everything you buy now will be out of date in 3 years. Here’s the example I used when I wanted to sell EqualLogic to my customers in the past.
So I always started wth a generic model that only covered the next 12 months. But due to it’s sublun tiering I could scale for same price towards performance or towards capacity. But guess what: I would never have been able to scale this to a petabyte. I would never have scaled this for 30 hypervisors. But hey: I would never need to go there! And even more; over 90% of my customers would never have outgrown an Equallogic cluster. And if they did, maybe a second cluster would have helped them and that wouldn’t really have been a big problem when you hit that level.
It’s good enough
So here’s my point. Why would I need a vBlock if my environment will never outgrow 4 Nutanix boxes? Why would I need a vCloud Suite (or the whole System Center stack) if my environment won’t outgrow my 50 Hyper-V VMs? And why would I need a petabyte-scalable-incumbents-array if my environment is 100TB big? Did you know you can easily put 1000 VDI’s on that single Tintri node? Guess what: that’s fine be me.
So now that we have that scalability issue tackled can we get back to business? Let me use that Tintri box again as I know that one best: although it’s a hybrid array (flash + hdd) it provides me 99% cache hit ratio out of that flash. Without even knowing what, there has to be innovation in the filesystem to get to that level. This alone makes this interesting enough to look at the product. And we haven’t even touched on that ease of management yet. The first time I saw the VM-centric way this system is designed was a revelation. I’m sorry for the others but I hadn’t seen that approach earlier. And please don’t come complaining that these are not global companies that would never have the support you need. Guess what: chances are big these guys have better support than the big boys out there.
You know who is eligible to buy those Tintris, Nimbles, Tegiles (and others not mentioned in this post)? The other 90%
External Link: you should definitely read “O ye, with little faith” by Sudheesh Nair, VP of Sales at Nutanix. It fits perfectly with this post.