And here is his storage predictions for 2010 and the year beyond

Storage Predictions for 2010 and Beyond

I’ll start with some easy predictions this time around. 

• FCoE will become available end-to-end, with major storage vendors supporting FCoE interfaces by the end of 2010. This should be an easy prediction, as the market is demanding FCoE given the potential cost and cabling savings. It takes longer for the design integration and testing of storage controllers, so it is no surprise that this takes longer than server or interconnect changes.
• PCIe 3.0 will make it to market, with availability in blades first. This is almost a given at this point.
• Multiple storage and server vendors will address the end-to-end data integrity problem. Now the predictions are getting tougher. There is definitely a market need given the documented problem of mis-corrected or undetected errors in the data path, and I believe that vendors will fill this market need in 2010. This will be the only major change in file system technology and the first major change in a long time.
• There will be market consolidation in the flash solid state drive (SSD) market in 2010. The number of vendors in the market is just too large for the size of the market. Some companies will either merge, get bought out, or disappear. The market for flash disk drives is only so big and it cannot support the large number of companies in the market even with the growth of the technology. And STEC (NASDAQ: STEC) will finally get some competition in the enterprise SSD market from the likes of Pliant and Seagate (NASDAQ: STX).
• Flash SSD usage will increase, with multiple RAID vendors and multiple controller card vendors providing better support, which means higher bandwidth and more IOPS in 2010.
• 10 Gigabit Ethernet will become the standard connectivity for almost all systems. Higher-end home PCs from Dell, HP and others will support this technology. Home routers from multiple vendors will have this support, likely before the end of the year (see Enterprise Technologies Will Change the Consumer PC Market).
• NFSv4.1 (pNFS) will enter the market with products from multiple vendors in 2010.
• 40 and 100 GbE will continue their march to product availability, with the potential for some interconnect between switches available in 2010, but with certain availability in 2011. For 40 GbE to be viable for the host side, dual port, will require PCIe 3.0 with at least 16 lanes, and at that PCIe 3.0 provides only 16 GB/sec of bandwidth, while 20 GB/sec is needed for full rate, full duplex operation. PCIe 3.0 is a must for host side connectivity.
• Last but not least: By the end of 2011, the cloud hype that you hear today will be greatly diminished. Clouds are good for some things, but just like the storage service providers and application server providers of the late 1990s or grid computing in the early part of this decade, clouds will meet a similar fate. SSPs and ASPs are providing services for some applications, but they are not going to solve all problems for all enterprises, as there is just not enough network bandwidth, the latencies for some applications are too high, and the security problem end-to-end has not be solved in a standard way (see Why Cloud Storage Use Could Be Limited in Enterprises). Besides, does anyone really see enterprises giving up control of their most critical data? I sure don’t.

In some perspective, it is more comparable to FCoE (Fiber Channel over Ethernet) than iSCSI. It export block devices (SATA HARD DISKS) over the network with a very high throughput when coupled with a quality Ethernet switch. A qaulity Ethernet switch can maximize throughput and minimize collisions through integrity checking and packet ordering.

In a typically implementation of AoE, you will need some dual-port Gig-E cards, a Gig-Ethernet switch and some disks.

#####
# Comparison of AoE vs iSCSI
################################

These are the following advantages of AoE over iSCSI

    * AoE is cheap and simpler software stack. The advantage of AoE is that you don’t have the overhead of translating ATA to SCSI then back to ATA if you are using ATA drives. So there is a performance pickup.
    * Server processing load for iSCSI is much higher than AoE for equivalent throughput. AoE can spare processing cycles. iSCSI requires TCP/IP and its requisite complexity.
    * AoE is not a routable protocol. Therefore it provides you inherent security.
    * AoE Ethernet frames are passed by standard switches.
    * AoE and iSCSI both have initiator support for Windows and Linux.

Disadvantages of AoE over iSCSI

    * If you need features such as encryption, routability and user-based access in the storage protocol, iSCSI is a better choice.
    * AoE is not much suitable for critical enterprise applications. AoE is not as scalable as iSCSI or Fiber Channel when you consider location i.e., with Fiber Channel and iSCSI, you can scale your storage throughout. This is primarily due to the inability of AoE to route AoE traffic.
    * ATA disks are not as reliable as their SCSI counterparts.