In-Depth

Spring Is Busting Out All Over

New technology blooms in the Storage garden

• By Jon William Toigo
• 04/07/2005

As the spring thaw begins , both flowers and weeds emerge from the frozen ground: so, too, with storage technology. For you data gardeners out there, here are a few budding developments you may want to lookout for.

First, Microsoft “Chimney” is supposed to be generally available soon. If you haven’t heard of Chimney, the implications of this fact may elude you. Essentially, Chimney represents a huge jump forward in TCP Offload Engine (TOE) architecture as an enabler of more efficient IP-based applications, including storage. The advance is not so much one of technological advancement as it is technology adoption.

Chimney will see TOE technology (already perfected in many respects by pioneer Alacritech, I am reminded) going mainstream. Expect to see motherboards appear in the market shortly with TOE built into their Gigabit Ethernet ports. Using the magic of Microsoft’s de facto standard-making capabilities, Chimey will create the necessary standard software stack to which TOE chipmakers can interface their wares in Windows servers. Comparable developments are underway in the UNIX and Linux worlds, at Sun Microsystems, and elsewhere.

One might argue that these developments signal that the days of conventional Host Bus Adapters (HBAs) are numbered and that TOE-enabled Network Interface Cards (TNICs) will shortly replace HBAs to provide combined LAN and iSCSI SAN support. Indeed, Alacritech predicted this development several years ago, .but they got their clocks royally cleaned by an entrenched group of HBA manufacturers who wanted to silence the upstart newcomer for making such a radical (though obvious) observation: the HBA emperor would soon experience a wardrobe malfunction.

Alacritech was right on, and they are now vindicated by lucrative OEM deals with brand-name vendors who are themselves scrambling to build iSCSI-enabled storage platforms to address the price sensitivities (though not necessarily the storage requirements) of the small- to medium-sized enterprise. By the time you read this , Alacritech will also have announced a boatload of TNIC products, propelling them into the forefront of the iSCSI connectivity game. Sometimes the good guys actually win…a round, at least.

A second potential ray of sunshine in your data garden is the advent of products based on the Serial-Attached SCSI (SAS) and Serial-Attached ATA (SATA) II protocols. First-generation SATA took the industry by storm last year, moving a lot of disk capacity out to consumers at rock-bottom prices. SAS and SATA II (the latter is a protocol refinement that enables SAS and SATA components to work and play well together) could well deliver the one-two punch to Fibre Channel in the very near future.

According to advocates, SAS costs a lot less than Fibre Channel and enables the connection of almost as many devices as FC along the internal backplane of the storage array. SAS is also fully compatible with SATA II, so you can build tiered storage inside a single array from both “enterprise class” dual-ported disk drives (SAS disks) and the more commodity-priced and single-ported SATA II disk.

With an external iSCSI connection, SAS/SATA II arrays might provide the means for ending, once and for all, the misapplication of FC fabrics to supporting applications that don’t absolutely need them (e.g., apps with extraordinarily high transaction rates). I’d be willing to guess that iSCSI, SAS, and SATA II could support about 80 to 90 percent of open systems application workloads in the market today.

One cloud in the sunny outlook on SAS/SATA II, however, comes from industry insiders at AMCC. They note that SATA II is being Balkanized by many vendors: instead of delivering the full suite of capabilities touted in the spec, including 3Gig “wire speed” support, native command queuing, and port replication, some vendors are shipping only a subset of these capabilities in their SATA II products. We are starting to hear product names that reflect this phenomenon: “SATA 3Gig Wide” and “SATA 1.1 Extensions.” These products do not deliver all of the bells and whistles we have been led to expect from SATA II. Caveat emptor.

The AMCC folks gave me something else to think about. They claim that SAS may not be as important as some in the industry would have us believe. With their current and upcoming SATA II implementations, AMCC claims it can deliver a compelling alternative to SAS through dual configurations of SATA II controllers and redundant SATA II disk that will deliver all of the performance and resiliency touted by the SAS guys but at a fraction of the price. They have promised to send us an array thusly configured so we can test it for ourselves. We’ll let you know what we find out.

Finally, be on the lookout for some great new technology from DataCore Software. We haven’t heard much in awhile from the Ft. Lauderdale, Florida storage software company, and rumors of its impending demise were running rampant only a year or so ago. To get the low down on what was going on at DataCore, I jumped in my roadster and traveled from the Gulf coast of Tampa Bay to the company’s digs on the Atlantic side of the state.

CEO George Teixiera and wild man CTO Ziya Azal—and the real powerhouse of the organization, Bettye Grant—convinced me that the trip was well worth the two tanks of gasoline it cost to visit them. First, the update: DataCore met the slowdown in the storage market head on in 2003, and the company went through a painful downsizing not unlike what I have seen at many Silicon Valley shops. The difference in DataCore’s case was that the management and employees bought out most of their investors and hunkered back down to work on some very interesting technology.

Within the next few months, look for them to deliver the first working implementation of virtualization over iSCSI, as well as some very cool I/O process expediting ware that will soon be turning a lot of heads. While they may be quieter than they have been in the past, they still demonstrate a passionate irreverence toward entrenched and institutionalized storage concepts that makes them out-of-the-box thinkers and innovators.

If their latest wares catch fire, we might begin to see an unraveling of the RAID controller value proposition. Who needs RAID if you can do copy-on-write to virtual volume sets? Again, it will be nice to see the good guys come back to win the day.

Next time, we’ll look at some of the weeds that you may want to pluck before they have a chance to overwhelm your storage garden. Until then, write me at [email protected].