Matthew Wallace at [1]AllThingsD wrote up a great article about how
  organizations employ a myriad of tactics to avoid the risks of shared
  storage environments, often inefficiently and ultimately self
  defeating:

    Massive overprovisioning of resources in clouds, dedicated storage
    platforms attached to shared compute platforms, dedicated shelves in
    shared storage platforms, or massive horizontal scaling are options
    used every day. They don't solve the problem - they avoid the
    problem, often at great expense or through significant architectural
    shifts.

  My take away from this article is to ask the right questions of your
  cloud storage provider or your storage infrastructure vendor to make
  sure you're not impacted by "Noisy Neighbors":

    For instance, does your CSP work with a storage vendor that offers
    guaranteed QoS on a storage platform? ... Cloud environments empower
    you with the business agility of service on demand and flexibility
    to respond to changing business needs rapidly. Adding resources for
    a time and then giving them up when they are no longer needed is a
    major benefit. While the advancement of cloud computing has made
    those accessible on the compute side, the storage side was left
    behind by the limitations of rotational disks and the inability to
    offer ironclad QoS guarantees.

    The power of a such a solution ... is not only in knowing that you
    can guarantee a certain number of IOPS on each volume, but to pair
    that with cloud environments to allow the business agility to burst
    as needed on the storage array the way that cloud environments offer
    that flexibility for compute.

    The rapid and automated provisioning world of the cloud demands that
    storage companies build APIs rich enough to control every aspect of
    an array. Building the user interface as a layer on top of the API
    is a demonstration of API and design maturity that shows a solution
    is future-proofed against demanding cloud orchestration
    requirements. Designing the solution to be linearly scalable without
    artificial breakpoints or step functions in performance keeps the
    provisioning and growth simple and reliable, shutting out the noisy
    neighbors once and for all.

  via [2]The Problem With Noisy Neighbors in the Cloud - Matthew Wallace
  - Voices - AllThingsD.
    __________________________________________________________________

  My original entry is here: [3][Storage] The Problem With Noisy
  Neighbors in the Cloud - Matthew Wallace - Voices - AllThingsD. It
  posted Tue, 26 Feb 2013 13:06:35 +0000.
  Filed under: technology, cloud, data center, shared storage,
  virtualization,

References

  1. http://allthingsd.com/20130225/the-problem-with-noisy-neighbors-in-the-cloud/
  2. http://allthingsd.com/20130225/the-problem-with-noisy-neighbors-in-the-cloud/
  3. https://www.prjorgensen.com/2013/02/26/sorage-the-problem-with-noisy-neighbors-in-the-cloud-matthew-wallace-voices-allthingsd/