No video this time! I felt this one would be easier to write down, be warned it is a long one, you may want to get that cup of coffee first!
Over the last 5 or so years I’ve worked at NetApp there’s been a constant theme to my presentations, they all focus on the value that ONTAP delivers upwards, enabling granular RTO and RPO using SnapShots and SnapRestore, integrating this into the applications with our SnapManager tools, very neat plugins like VSC for VMWare and providing DR and Business Continuity with SnapMirror and MetroCluster. I’ve even taken this to the level of demonstrating how capabilities such as FlexClone, when put into the hands of the people that really need it, can dramatically reduce test / development cycles and can even enable the adoption of Agile development methods.
These are the ‘Visible’ capabilities, they’re the tools and technologies that people interact with, that people call on when they have a need. They are also the capabilities that we are starting to see built into the next generation of applications, Exchange for example, while limited in terms of its Data Protection tools does do a very good job of DR using its inbuilt Clustering capabilities. I think this is where we’ll see Software Defined Storage (SDS), or Virtual storage really begin to manifest, it will start to pool storage capacities and present these up into the hypervisor (Software Defined Server) layer with increasingly more capabilities and application integrations.
What I really don’t spend much time talking about (and I probably should) is the ‘under the hood’ capabilities. For over 20 years now we’ve been working on the technologies that hopefully you’ll never see and in most cases probably aren’t really even aware of. When you look at the anatomy of a storage array and how chunks of data are received from hundreds or thousands of users in parallel, intelligence inside the array then makes decisions as to how these millions of blocks should be assembled into RAID stripes and then laid down across the disks at the backend, all the while these blocks continue to be accessed. Imagine turning up to watch your team play, 10s of thousands of you all streaming into the stadium at pretty much the same time, then someone calling out for individuals, rows or complete blocks to stand up and head back out again, can you imagine the chaos? Yet this is what we do and we try to bring a sense of order and safety to it at the same time, because it also has to happen really fast!
Add to this that we’re working with mechanical drives which occasionally fail, which is okay when it’s an obvious failure but what about when it’s just a media error? Did that one block in the 100’s of thousands you just wrote actually make it to the disk? Who checked? What if it didn’t? How do you recover it? How do you make sure that everything that needs to know, knows exactly where that block that was just recovered now lives, it’s certainly not where it was before.
So why do I call these the ‘Visible’ and ‘Under the hood’ technologies?
I’m a car guy, I’m also an engineer, so its probably no surprise to you that when I think about NetApp the picture in my mind is a car. There are some very visible features and capabilities, from the way its styled, to the myriad of controls inside the vehicle that make the whole process of driving and navigating that much more simple, these are the things that are valuable to the driver that make his life more comfortable, easier and even faster, and these are the things that are typically easy to explain because they’re something the driver will interact with on a daily basis.
Then there are the ‘Under the hood’ technologies, the engine, now we all probably know how many cylinders our car has, probably also the engine capacity, but how much more do we really understand? How much more than that are we ever really told? Out of all of the components in a car the engine is without doubt the most sophisticated. Crankshaft and camshaft moving at many thousands of RPM, cylinders moving back and forth inside the chambers at many times per second and all of this engineered over decades to run without fail for 100s of thousands of miles, typically with just a few warning lights to advise us to top up the oil or the water.
The most important and complex part of any system is often the part you don’t see!
For me this has many parallels with a storage array – we know how many disks and shelves we have, we know what capacity they deliver, but how much else? We expect them just to work, for thousands and thousands of hours spinning at 10 or 15,000 rpm, the engineering / expertise that has gone into making this happen simply cannot be under stated, and we also expect warning lights to alert us to potential failures before they make us pull over to the hard shoulder.
To take this a step further, we have seen the emergence of Hybrid vehicles, mechanical engines coupled with electrical ones to work in harmony with each other depending on the way the car is being driven. In the world of storage we also have the Hybrid array, utilising mechanical disks in conjunction with electrical ones (Flash) depending on the way that the storage array is being driven by the applications. And now we’re seeing the emergence of the ‘All Electric car’ and ‘All Flash Array’ where many of the traditional mechanical technologies are being replaced altogether, as with all newer technologies they typically command a premium and they’re not right for everyone, capacity tends to be less so if you’re a bit of a road warrior then the all Electric car probably isn’t right for you yet, likewise if you’re storing large capacities of data then the All Flash Array also isn’t probably the right technology for you, yet! And make no mistake, they still fail, they just fail in a different way.
Recently I’ve had a few customers talk to me about building their own storage environments, using VMWare VSAN or Microsoft Storage Spaces on top of commodity storage hardware, I personally don’t believe these tools have the scale or maturity today, but in time they will become an option for some environments. And whilst I agree that these tools will enable more choice, and will enable more commodity type storage to be adopted, it absolutely does not mean that you can use simply anything behind them, as I mentioned before, the engine is the foundation, the most complex part of the entire vehicle and if it isn’t robust or reliable then whatever you surround it with IS going to fail at some point, it’s just a matter of when, how and worse still how long it happens for before you even notice.
People thought server vendors were doomed when VMWare launched into the market, that companies would be able to use any server behind it, it has had an impact, I’m not going to analyse that here, but at the same time, the very same time that companies were predicting the demise of servers, Cisco launched themselves into the Market with UCS and have become one of the largest server vendors in the world today, there really are some huge benefits to using Cisco’s UCS hardware as the server / network foundation.
So how is NetApp helping you to navigate this road?
Our FAS Hybrid arrays using Disk and Flash at different levels running cDOT, have extreme efficiency, unified architecture, NDO capabilities and are leading the way in what you should expect from a shared Storage infrastructure. Link this with Cisco UCS as FlexPod and it’s your perfect vehicle, the best engine completely integrated into the best chassis and body.
Maybe you do want to ‘build it yourself’ and use a layer of software based storage capability, such as VSAN or Storage Spaces, then you should consider our E Series array, a high performance, high density and rock solid cost effective engine to take care of those ‘Under the hood’ activities, and in its ‘All Flash’ version, the EF550, it provides a level of performance and reliability that offers a very real alternative to traditional Tier 1 storage arrays.
Or maybe you want to do want to design it even more yourself and look at using server based storage, in which case let me introduce you to ONTAP Edge, our software based solution which incorporates our engineering know how from over 20 years of test, development and innovation
The market is changing, people will deploy storage in many different ways in the future, I believe NetApp is extremely well positioned for this, whether you want a full vehicle or just a powerful engine, from a visible look and feel straight through to under the hood innovation.