
|

|

|

|

 |
| Tech Update
|
 |
Why SAN and NAS will converge
|
By Sean Derrington
Meta Group
March 14, 2002
| Provided by |  |
|
|
 |
|
META Trend: Fibre Channel (FC) storage network deployments will continue unabated through 2005/06, and will not be impacted by emerging technologies before 2004 (e.g., iSCSI and InfiniBand as a storage interconnect). As iSCSI products mature, they will complement FC for specific purposes (e.g., low-cost direct attached storage, long-haul data replication). InfiniBand will be used primarily for Intel-based server CPU communication and interface to TCP/IP (LAN) and FC (SAN) networks.
There has been much confusion in the market about current and future capabilities, differences, and limitations of NAS and SANs. We believe this will continue as the architectures collide and, by 2005/06, a truly combined storage architecture becomes viable. Through 2005, about 80% of networked storage (storage that is attached to multiple servers/clients) will communicate via block/channel protocols (SCSI-3 commands) and approximately 20% will be communicating via file protocols (NSF, CIFS, HTTP). From a technical perspective, information requests are satisfied by either blocks (SAN) or files (NAS), and with both architectures the "1s and 0s" are ultimately written to mechanical disks (irrespective of the device interconnect--ATA, Ultra SCSI, Serial Storage Architecture [SSA], or Fibre Channel [FC]) via SCSI commands, the difference is where the file requests are translated into block requests.
| [an error occurred while processing this directive] | Although we view SANs as a set of principles by which data storage and backup/recovery traffic are offloaded from the application network, SAN disk storage refers to the capability of a redundant array of independent disks (RAID) subsystem to communicate with servers/clients via SCSI commands (irrespective of the transport) and not via TCP/IP (file systems) and Ethernet. EMC was the first vendor to combine the concepts of NAS access (appliances satisfying file requests) with SAN (channel) delivery based on thresholds (size, performance) determined by IT organizations (ITOs) with the introduction of HighRoad. Others have followed suit, and by 2003/04 we believe there will be more offerings of a similar ilk. However, ITOs still must understand the implications of vendors touting such capabilities as NAS appliances (a.k.a. NAS heads or dedicated file servers), which alter storage management operations.
Although some things are simplified (e.g., file sharing, administration, installation), other things become more complicated (e.g., backup/recovery, virus scanning, scalability, breadth of applicability), and regardless (of a dedicated or combined architecture), the translation must take place somewhere; this understanding (the benefits/drawbacks) is critical for successful implementations.
NAS niche? Survival of the flexible?
Given that 15+ vendors have entered the much hyped NAS market (either developing their own technology or through OEM relationships), we believe vendors that only offer a NAS solution and do not address the larger, and arguably mission-critical, market of channel-attached storage (SAN) will be eliminated by 2004 (likely 80% of these). Once again, EMC pioneered the notion of sharing physical storage hardware (Symmetrix) with NAS appliances (Celerra) and general-purpose servers (e.g., AIX, HP-UX, Solaris). Subsequently, HDS has followed suit with its Network Storage Solutions partnership, which has been marginally successful, at best.
 |
 |
|
|
|
 
[an error occurred while processing this directive]
![]() |
|
[an error occurred while processing this directive] |
![]() |
 |
![]() |

[an error occurred while processing this directive]

|

|

[an error occurred while processing this directive]
|
[an error occurred while processing this directive] |
 |
[an error occurred while processing this directive]

|

|

|

|