FiberChannel storage area networks are a great advancement when it comes to mass storage frames. High performance (100 MB/s) is realized by switching the fiberchannel frames/data instead of using a shared bus architecture like SCSI. This way very large disk frames to the tune of dozens are terabytes and hundreds of volumes can be housed in a single managed device.
My rub is the cost of fiberchannel switches especially since this kind of pooling of resources using a high-speed LAN is nothing new. FC switches can run $2000 per port not to mention the expense of the blades in the disk frame. Gigabit Ethernet on the other hand is around $500 per port using fiber.
Seems to reason that with Gigabit ethernet so mature you could use the higer performing ethernet switches to replace or displace the FC switches. That is where SCSI over IP comes to play.
So, who will win the upcoming battle for SANs? 10 Gigabit ethernet is already out and sure to drive down ethernet costs yet again. To me iSCSI is a clear winner considering the existing network management tools available today, higher performance, and the ability to scale with IP.
Should be an interesting two-three years for stroage.
My rub is the cost of fiberchannel switches especially since this kind of pooling of resources using a high-speed LAN is nothing new. FC switches can run $2000 per port not to mention the expense of the blades in the disk frame. Gigabit Ethernet on the other hand is around $500 per port using fiber.
Seems to reason that with Gigabit ethernet so mature you could use the higer performing ethernet switches to replace or displace the FC switches. That is where SCSI over IP comes to play.
So, who will win the upcoming battle for SANs? 10 Gigabit ethernet is already out and sure to drive down ethernet costs yet again. To me iSCSI is a clear winner considering the existing network management tools available today, higher performance, and the ability to scale with IP.
Should be an interesting two-three years for stroage.
