Yesterday a co-worker – in fact my new boss – sent an email to my fellow PM Steve and I about Server SAN that he found on Wikibon. You can see the links here and here. I think they have a bunch of great points (BTW, Server SAN is when you use local disk in your servers to provide shared storage). In particular they talk about how Amazon or Facebook have already moved down this path and are using great huge environments of Server SAN. Which is good for them, in fact likely what makes what they do possible but it is not as interesting for many of the customers and environments I know of. I agree with Wikibon in that Server SAN is going to disrupt the storage market. It is based on the simple reasons of simplicity and ease of implementation and management, with the usage the same or better after that. This means Server SAN in theory will be successful as it is simple to manage and consume, and easy to implement and yet while different it is consumed the same as always. Plus it has advantages today as you can treat some workloads different than others. We do have an example of this that I think highlights the possibilities and it is VMware and their VSAN product.
In the early days of VMware’s VSAN I was a big part of it. My VP at the time assigned me to it full time – which was unheard of at the time and he explained it as he thought that VSAN would be more disruptive than virtualization over time. I think he was a little melodramatic but he had the right idea. VSAN is way better than people realize. It can handle a lot of things going wrong and keep working. It is more forgiving, and faster than people expect. It lets people think of applications and service levels easily and smooth. It will take a while but VSAN and things that come after it will push out traditional SANS from the majority of customers who have them today. They are easy, and simple to use. That resonates with customers in a big way, especially when they find it a decent price and that it works. I think many customers over time will not be able to justify a traditional SAN. So smaller storage companies, and even big storage companies will be impacted – in particular if they cannot adapt.
VSAN, and Server SAN in general is in its early days. It doesn’t do replication for example, or have agents to help with application consistency during replication. But it doesn’t need all of that right now, and in fact VMware has the ability to add what is needed – especially if EMC lets VMware do what is right for customers.
What I am talking about has started. It will take a number of years, and it can in fact be delayed and confused by vendors but it will happen. It will not happen due to big data or VDI, but rather good prices, easy admin / implementation and because it will fit with customers really well.
Want to learn more about VSAN – or maybe get it going in the lab? You should and you will understand more about how I feel and what I have written about. You can find what you need here, and here – in fact, much more than you need!
As always, comments are welcome!
=== END ===