Storage Networking: The Next Big Thing?
Peer-to-peer networking....supercomputing...distributed computing...
Whatever you care to call it, the idea of getting groups of PCs to work together in harmony to complete a task is a thing of beauty to people for different reasons. For the tech-heads, the thrill seems to be in the process of interweaving the computers -- getting them to actually function like the Waltons not the Bundys.
Then Napster came along some time ago and peer-to-peer networking has become not only lucrative to techsters, but sexy and widely popular among music lovers -- and it's the singular, core technology that has sparked hours of contentious debates and legal brouhahas.
Well, other things exist that work harmoniously, too, but without causing all of that controversy. For one, Storage Area Networks (SANs) could be the next big thing, and banking on that theory are hardware networking manufacturers such as Cisco Systems Inc. and Brocade Communications Systems, both of whom plan to announce important new products in the sector.
These companies and dozens of other vendors will take to Palm Desert, Calif. for Storage Networking World Spring 2001 next week, April 9-11. ITIS Services, AT&T Solutions, BMC Software, CNT, Compaq StorageWorks, Computer Associates, DataCore Software, DLTtape, EDS, EMC and Exabyte are just a smidgen of the 90-odd sponsors to take up some space at what is anticipated to be the hottest show of its kind to date.
SAN-k You Very Much!
So, what the heck is a SAN? A SAN is a high-speed network, consisting of servers and storage. Its purpose is to handle large amounts of data traffic between servers and storage devices, without reducing the bandwidth of LAN/WAN -- the Local Area Network and/or Wide Area Network.
Don't confuse SANs with LANs though. A LAN sends smaller packets of data with increased communication overhead, reducing bandwidth. A SAN uses storage protocols (SCSI), giving it the ability to transmit larger pieces of data with reduced overhead and higher bandwidth.
Operating behind the server, SANs are often connected by a super-fast transmission technology known as the Fibre Channel. A SAN consists of a communication infrastructure, which provides physical connections, and a management layer, which organizes the connections, storage elements and computer systems.
And like the much-ballyhooed 3G, XML, and numerous other technologies, firms are ramping up SAN solutions to the market; the SAN is an invention of necessity. Traditional ways of connecting servers and storage no longer cut it because they are too slow. Moreover, legacy protocols are no longer suitable for handling large amounts of data. Most subscribers to the network-is- almighty theory feel that as computing environments move from a server-centric to data-centric model, access to shared data resources becomes critical. Think of SANs as the enabler in the crisis situation.
So SANs are good things right? Maybe, but they could be better. While firms have long praised SANs as a reprieve from Internet Protocol (IP), most have come to find that they are stymied by impractical standards that have meant that network switches and disk storage systems in servers have been unable to work with each other unless strenuously tested. Still, many insiders believe IP networks will lay a new foundation for SANs and will ultimately replace fiber channel.
Cisco's Storage Products Still Anybody's Guess
Solving such communication breakdown problems, it seems, is what gatherings like Storage Networking World Spring 2001 are intended to do. In a research note on the event, Goldman Sachs' Laura Conigliaro said that while there will be the requisite hype (no different than any other show, really), she expects progress reports on new switches and IP storage, particularly from Cisco and Brocade.
Steeped in ambiguity, Conigliaro said Cisco "will announce products to make SANs work better with the broader network."
San Jose, Calif.-based Cisco Thursday issued an advisory about products it would announce next week but did not offer a preview of what was in store. Cisco spokesperson Jon Noh told InternetNews.com Friday the company would not raise the curtain any sooner than a 1 p.m. PDT press conference on Monday to be "equitable" to everyone. Interestingly, Noh said that in addition to slews of calls from the press, fellow vendors called to inquire about Cisco's product announcements, too.
Also, Goldman Sachs refused to confirm that they were briefed and would not speculate further. So the "Cisco Storage Networking Initiative" as the firm billed it, remains shrouded in secrecy.
Just a couple of days ago, the Cisco said its optical switching business had folded, causing some analysts, like MetaMarkets.com Dave Nadig, to estimate that the company's acquisition of Monterey Networks in August 1999 was a waste.
Nadig called it a failure, and one that competition, especially optical switch leader Ciena, could look favorably upon. Obviously, Cisco (NASDAQ:CSCO) could use a positive reception Monday. Cisco shed more than a buck Friday and was trading at $13.62 as of press time.
But Conigliaro was decidedly more candid about Brocade, which she said could possibly unveil a director-like switch at the show.
"While Brocade's network switch is not meant to be an exact replica of directors on the market today, it is likely to have numerous high-end features that should sit well in the high-end enterprise space," Conigliaro wrote. "This product will use a high-performance backbone and Brocade's new 2 Gbit/second ASIC that gives it significant performance enhancements over its current line of products."
Conigliaro also detailed an important development to the storage industry; 2 Gbit per second fiber channel switches, which will supposedly give fiber channel SANs a leg up on IP networks, will roll out by the end of 2001.
As for the holistic view, Conigliaro said much work is needed for this pioneering technology.
"Products that have been announced still require much testing and qualification. which could take several quarters, so it is unlikely to see much in the way of production shipments until H2," she wrote. "Many in the industry, including some companies with a neutral bias continue to believe that IP-only storage solutions will not be data-center capable until at least a couple of years from now."
Lisa Beldean, technical director of customer solutions and product development of storage networking consulting firm ITIS Services, of Norwalk, Conn., completely agreed with Conigliario's assessment.
"I agree that IP storage is far from providing the performance of fibre channel SANs," Beldean told InternetNews.com. "IP in general can not compete with fibre channel efficiencies, regardless of the protocol it wraps inside it. However, cost is still a huge obstacle for SAN adoption and keeps lower-cost IP storage in the limelight. It was a compelling argument that if you are going to build a secondary network anyway, why not use the ethernet that we are familiar with, and, ethernet had gigabit speeds while fibre channel only had 100 MBps. Price/performance for the ethernet solution was still attractive to many. Now that fibrechannel is leap-frogging with 2 GB speeds, ethernet/IP storage is behind in the game again."
Beldean also said confusion among consumers who need SAN infrastructures is the biggest challenge facing progress in the sector.
"The biggest problem now is confusion," Beldean explained. "Most consumers do not know the true differences between SAN, NAS, IP Storage and Infiniband. Most customers I speak with today still do not know that SANs are viable technology actually in production in companies -- not just labs somewhere. If you look at all of the manufacturer claims about their own products and warnings about heterogeneous environments, it's no wonder consumers are confused. There are no really good free sources on the Web that can independently compare these technologies and identify what's best. IT all depends on many factors in the customer's environment as to which direction they should head. Without an in-depth knowldege of all of them, customers can (and have) made costly mistakes on infrastucture investments."
Well, what then, is the reason for any hullabaloo, which is no doubt underscored by Cisco's secrecy, if the world will not see viable IP-based storage products within the next year?
Two words: market potential. While most research firms vary in their projections, nearly all agree that, like many other technologies knighted for fast-track acceleration, the market will bloom. Dataquest thinks the SAN marketplace will grow 89 percent compounded annually between 1999 and 2003; IDC says SANs are expected to grow ten-fold by 2002. And those may be conservative numbers.
Obviously, the storage industry is hardly mature. The world may know by the end of the conference next Wednesday whether it has gotten any older, pending certain vendor announcements.