So what's all this buzz around object based storage?

Object-Based Storage Device 

Object-Based Storage Devices (OSD) enable the creation of self-managed, heterogeneous, shared storage by moving low-level storage functions into the storage device itself and accessing the device through a standard object interface rather than a traditional block-based interface such as SCSI or IDE.  

Based on T10 OSD Specification

Intelligent storage increases horizontal scale

OSD Defined- SCSi Object Based Storage, peripheral deice type defined as OSD.

A storage object is a logical collection of bytes on a storage device with well known methods for access, attributes describing the characteristics of the data and security policies.

The difference between OSD and a block based device is the device interface. NOT THE PHYSICAL MEDIA BLOCK BASED DATA ACCESS -based on the physical block address.

2 components:

1.User component-presents logical data structures to the user applications

2.Storage component-manages the storage device manages the block meta data

Having these 2 separate components makes data sharing (sharing data blocks) between hosts very difficult.

BLOCK BASED DATA ACCESS -based on the physical block address:

Store( write block number) | Retrieve(read block number)

Devices relatively unintelligent and externally managed

Storage component manages meta data

FILE BASED DATA ACCESS:

Store(File) | Retrieve(Directory)

OBJECT BASED DATA ACCESS -based on the physical block address:

Store(DATA -->OID)| Retrieve(OID--> DATA )

OSD devices are intelligent and self managed

OSD devices are aware of the storage applications they serve

The most immediate effect of OSD is the offloading of space management. (Allocation and tracking of used & free blocks)

The interface OSD storage uses is very similar to a FILESYSTEM.

OSD manages the physical layout of the data (self management). Objects are created, deleted, read and written just like files are today With OSD management is offloaded to the storage device rather than the storage application. With OSD, in offloading the metadata to the storage device objects removes the dependency between metadata and the storage application.

With the metadata offloaded storage applications can now store their data structures as single objects rather than a collection of blocks.

OSD devices enables the creation of self-managed, heterogeneous, shared storage by moving low level storage functions into the storage device itself, rather that externally managed in block evices, and by acessing the device thru standard object interface rather than the block based interface such as SCSI or IDE. Self management benefits SAN and DAS.

Because the device can treat objects individually its easier to set security policies on a per object basis.

THE IMMMEDIATE BENEFIT OF OSD IS CROSS PLATFORM DATA SHARING IN A SAN.

THERE WOULD BE LITTLE BENEFIT WITH DIRECT ATTACHED STORAGE(DAS)

Posted at 09:08PM Nov 29, 2007 by gwaterson in Storage  | 

Storage Performance Numbers

I wonder why just about everyone except EMC posts results to Storage Performance Council. What are they afraid of?

Posted at 05:45PM Nov 20, 2007 by gwaterson in Storage  | 

2007 DMA Conference

I just returned from the annual Sun Data Management Ambassadors conference, a gathering of 100 of the best international customer facing storage engineers at Sun. 6 days of intensive engineering, roadmap and executive presentations. All I can say is WOW, fasten your seat belts and clear some rack space, we're in for a wild and market disrupting ride for the future of storage at Sun. Year after year hearing where we're headed with our storage platforms is an exciting adventure and this year is the most exciting plans I've seen. More to come on this subject as we move forward.

I've gotten real sick of sour grapes from naysayers about our storage. It's time we see the road littered with out competitors products. AND It sure will be!

Every storage sales call I go on my goals are simple: BEAT THE COMPETITION. How can I say this? Take a look at ZFS a brilliant file system packed full of features.

No more volume managers necessary.

Take a look at what we're doing with our Open Solaris Storage Initiative.

See all about x4500 Thumper and our 5800 Honeycomb

ZFS, x4500, and where we're going has changed our rerality!

We're open THEY ARE NOT!

From the low-end of our disk storage offerings to the high-end we have better products than any of our competitors (PERIOD!)

Posted at 10:51AM Nov 20, 2007 by gwaterson in Storage  |