The converged storage

Nowadays we speak a lot about new features, roadmaps, technical details about dedupe on primary storage or automated tiering and so on, but what is going to happen in the next few years?

Each vendor is trying to explain why its solution is better than others and how they have the ability to do more with less, support your explosive growth and simplify your data storage life (uhm!)
Small and innovative vendors have the technology leadership while big ones are running after them but with a lot of money to copy or buy good ideas!
In the last years we have seen a convergence in all about servers with the result of x86 and Virtualization predominance. Hardware for servers is no longer important: it’s a commodity! Yes, you may prefer IBM because it’s black painted or Dell for the leds on the front panel or HP because they are the #1 in sales or Cisco for something else but you can change the vendor whatever you want! The x86 race is no more about features and capabilities but about quality, support and services!

Other pieces of the infrastructure are collapsing in a converged layer and FCoE is the most famous example. We need to take care about the first hints: the storage controllers are already converging on x86, hard disks are standard, protocols and connection links are standard… someone is talking about Virtual Storage Appliances.

What will happen next? Will storage become a commodity in few years?

  • http://topsy.com/www.cinetica.it/2010/06/10/the-converged-storage/?utm_source=pingback&utm_campaign=L2 Tweets that mention The converged storage « Cinetica — Topsy.com

    [...] This post was mentioned on Twitter by unix player, Enrico Signoretti. Enrico Signoretti said: New blog post: The converged storage http://bit.ly/a0hzkA [...]

  • http://twitter.com/BasRaayman Bas Raayman

    I’m not entirely sure about your last question, since my answer would be that storage already is a commodity. When I look at Wikipedia’s definition (http://en.wikipedia.org/wiki/Commodity):

    “A commodity is a good for which there is demand, but which is supplied without qualitative differentiation across a market.”

    then I would say that this is already true for most storage environments. It’s easy to find a storage solution now that uses features such as thin/virtual provisioning, alternatives can be found to do wide striping, and storage tiering (although implemented differently, the results seem to be more or less the same) are already a commodity. Using virtualized controllers to connect to my storage makes it even easier to view my storage as a commodity, because it really doesn’t matter what kind of array I connect in the back end (at least, in theory).

  • Enrico Signoretti

    Bas,
    great point here, but if you see this post from the vendor’s point of view probably you’ll comment the opposite way, :-)

  • http://twitter.com/BasRaayman Bas Raayman

    Yes, but that’s simply not interesting. Sorry to put it that way, but those vendors exist because there is a customer need for such solutions, or an alternative for something. A vendor doesn’t want commodity, a vendor want’s to see a customer where his own products are the commodity. Any other choices, and something being a commodity are the result of customers having a choice and vendors having to implement a similar function. As soon as the majority implemented such a function, I would say that we are talking about this function being a commodity. Things like TP/VP, Zero Block Reclaim are funtions that are (turning in to) commodity.

  • http://twitter.com/BasRaayman Bas Raayman

    I'm not entirely sure about your last question, since my answer would be that storage already is a commodity. When I look at Wikipedia's definition (http://en.wikipedia.org/wiki/Commodity):

    “A commodity is a good for which there is demand, but which is supplied without qualitative differentiation across a market.”

    then I would say that this is already true for most storage environments. It's easy to find a storage solution now that uses features such as thin/virtual provisioning, alternatives can be found to do wide striping, and storage tiering (although implemented differently, the results seem to be more or less the same) are already a commodity. Using virtualized controllers to connect to my storage makes it even easier to view my storage as a commodity, because it really doesn't matter what kind of array I connect in the back end (at least, in theory).

  • esignoretti

    Bas,
    great point here, but if you see this post from the vendor's point of view probably you'll comment the opposite way, :-)

  • http://twitter.com/BasRaayman Bas Raayman

    Yes, but that's simply not interesting. Sorry to put it that way, but those vendors exist because there is a customer need for such solutions, or an alternative for something. A vendor doesn't want commodity, a vendor want's to see a customer where his own products are the commodity. Any other choices, and something being a commodity are the result of customers having a choice and vendors having to implement a similar function. As soon as the majority implemented such a function, I would say that we are talking about this function being a commodity. Things like TP/VP, Zero Block Reclaim are funtions that are (turning in to) commodity.

blog comments powered by Disqus