The Object-based Storage Device specification, announced by the Storage Networking Industry Association earlier this month, could potentially bring about vast, fundamental changes in the way we use and manage storage over the next few years.
The goal of the OSD specification is to move storage management functions away from file systems and databases and onto the storage devices themselves. By making storage hardware more intelligent, the OSD specification working group—composed of storage vendors—believes it will be able to improve end-to-end security, increase performance scalability and allow broad access to data by multiple clients.
Adding to these lofty goals, the OSDs Technical Work Group believes its specification will also make it easier for IT managers to use storage hardware from numerous vendors.
eWEEK Labs believes the emergence of OSDs will be good news for IT managers because these devices will help streamline storage management, which has grown out of control and become a big financial drain during the past few years.
However, like any technical revolution, it will take a fair amount of work for early-adopting IT managers and hardware and software vendors to iron out all the kinks.
Currently, IT managers have two storage options: They can choose high-performance storage solutions such as DAS (direct-attached storage) and SANs (storage area networks), both of which give servers and applications direct block-level access to storage but are not designed with data sharing in mind. Alternatively, they can choose NAS (network-attached storage) solutions, which have file-sharing capabilities to control access rights to data and facilitate sharing but usually operate at slower speeds.
|
Objects bridge the gap between blocks and files, providing IT managers with a third storage option that allows direct access to storage devices while still providing the security and sharing capabilities of NAS and file servers.
OSDs will be able to identify data as it enters the storage system and create metadata (which will have information on such things as file and user attributes) to classify them as objects.
The standards the OSD group is working on now will eventually allow a single namespace to be created so heterogeneous storage devices can communicate with one another and recognize data objects.
By using multiple data paths that run from the initiators to the OSDs in a network, IT managers will be able to create storage infrastructures that are scalable in size and performance.
In data object format, the OSDs will be able to decide how the object will be stored and what quality-of-service level it will need.
Storage vendors believe OSDs will be self-managing and self-healing in a few years. This is important because in a few years IT managers will no doubt have bigger storage management headaches.
By leveraging metadata, the OSD group can create policies to ensure less important data objects are automatically put onto standard storage devices, while performance-sensitive objects are placed on the fastest devices available. Whats more, all of this will be done at the storage-device level without human intervention.
In addition, OSDs will be able to expand capacity on demand, based on application requests, which should make day-to-day storage management chores significantly easier.
Since virtually every storage vendor in the market is working toward creating information life-cycle management solutions, OSDs should become popular because they will make it easier to automate data migration processes than current solutions do.
In the future, objects will probably have metadata in them to determine retention and expiration dates for data as well as information on usage patterns. This will allow storage devices to “prefetch” data before it actually is requested, shrinking user wait times.
Make no mistake about it, however: The move toward OSD will not be easy. It could take several years for storage vendors to harness the full potential of OSDs. On the hardware side, vendors will have to add intelligence to their storage devices to allow them to identify and manage the data stored on them. Applications and file systems will also have to be rewritten to work with OSDs.
For NAS vendors, CFIS (Common Internet File System) and NFS (Network File System) will have to be modified to allow direct access to OSDs.
Senior Analyst Henry Baltazar can be reached at henry_baltazar@ziffdavis.com.
Be sure to add our eWEEK.com storage news feed to your RSS newsreader or My Yahoo page: