Sprint Stretches Storage Over IP

Sprint Corp. (NYSE: FON) has completed a disaster recovery demo that enables data replication over a wide-area IP network, which it claims it can turn up as a service any time (see Sprint Readies DR Service).

What it doesn't know, is how much to charge for the service yet or when it might formally introduce it. The usual time frame for a new service to hit the streets once it's been approved in Sprint's Concept Realization Center [ed. note: a fancy term for R&D lab] is a year to 18 months. "We are still working on the business case around it," says Sprint spokesman Jeff Chaltas.

"This service could enable customers such as banks to replicate their data at extremely remote locations using their existing IP connections," says Oliver Valente, VP of technology development at Sprint. "It’s a huge cost saving over current methods."

Today, disaster recovery offerings from the likes of AT&T Corp. (NYSE: T) and other carriers link up data centers using a whole fiber or a whole wavelength rather than simply making use of existing IP infrastructure, which is much cheaper. “The last mile to the customer no longer has to be fiber, it can be FCIP [Fibre Channel over IP], which is significantly less expensive than DWDM,” says Audrey Harman, manager of the technical staff at the Sprint Concept Realization Center.

The demonstration was conducted using an FCIP connection, a protocol that's being standardized by the Internet Engineering Task Force (IETF), which encapsulates Fibre Channel storage traffic within IP packets for connectivity over wide-area networks.

The trial transferred a 125-Gbyte Oracle Corp. (Nasdaq: ORCL) database from a Sprint lab located in Overland Park, Kan. Optical crossconnect devices at Sprint's Burlingame, Calif., lab created a continuous loop back to the Overland Park lab, simulating a point-to-point FCIP connection spanning 3,600 miles.

According to Harman, the Oracle file was transferred one way (1,800 miles) in 3 hours, 8 minutes.

An alternative way of encapsulating storage over IP -- iSCSI -- has been tested by others over even longer distances (see iSCSI Travels 7,500 Miles). Sprint checked out iSCSI itself but decided it was too early in the adoption cycle for this technology. “Right now storage arrays do not support iSCSI and the standard has only just been ratified,” says Harman (see iSCSI Gets Go-Ahead).

The equipment used to create the demo -- private Sprint circuits, Cisco Systems Inc. (Nasdaq: CSCO) MDS 9000 SAN switches and MDS 9000 IP Storage Services modules, and Hitachi Data Systems (HDS) Freedom Storage Lightning 9900 V Series systems running Hitachi TrueCopy data replication software -- enabled link utilization of 100 percent. “Most data transfers don’t fill the pipe completely, so it takes longer to send the data,” Harman notes.

Cisco and HDS got first dibs on the deal as both are already strategic partners of Sprint. Harman says Sprint is currently inviting all the major storage switch and array vendors to join the demo (see Sprint Puts Cisco to Test).

The timing of this announcement is pertinent in the light of recent government regulations around data backup and recovery for financial institutions. Although the regs do not mandate the distance between remote sites, U.S. federal agencies are requiring "firms that play significant roles in critical financial markets" to recover clearing and settlement activities as soon as possible. They must recover operations "within the business day on which a disruption occurs with the overall goal of achieving recovery and resumption within two hours after an event." Any entity that fits the description of a "core clearing and settlement organization" must meet these standards by the end of 2004 (see Feds Set DR Regulations).

The Feds also require that a backup facility be fully staffed in the event of a disaster, which could be a financial challenge for some organizations. “Sprint’s offering means that companies can backup to a remote location where they already have resources, taking out the cost involved in building new data centers for disaster recovery,” says Frank Barbetta, analyst with Probe Research Inc..

Analysts estimate there are at least a hundred financial organizations in the U.S. that will need to upgrade their networks to be in line with the new regulations. And storage-over-IP contracts are beginning crop up. Cisco recently won a deal with Euronext LIFFE, a financial futures exchange in Europe (see Cisco Touts Storage-Over-IP Project).

— Jo Maitland, Senior Editor, Boardwatch

whyiswhy 12/4/2012 | 11:47:36 PM
re: Sprint Stretches Storage Over IP FCIP SAN as a business worth exactly the same as remote web hosting is worth...just about zilch.

With a standards based business, there is no competitive edge except finding the cheapest hardware, locating it in the cheapest wharehouse, with low cost electricity, and getting the cheapest sysops you can find to maintain it. Probably physically located in India, using Linux servers.

Sprint can do it because it has scale. But so can anyone else with scale.
mordecai 12/4/2012 | 11:47:21 PM
re: Sprint Stretches Storage Over IP what follows is a rant. don't read it if you don't want to.

IMHO, this press release is complete crap.

you should be ashamed of yourselves for a useless hero experiment where you give IP a perfect network connection to work into.

You know that this exact deployment can be done for less $$ with a CNT USD on a DS-3. been done like that for years and years. And it works in a real network, unlike 'real IP'.

You are not limited to dark fibre or lambdas...

this is essentially a private line nework with an ip conversion on the ends.... this does not address the key DEEPLY ROOTED FUNDAMENTAL ISSUES that make IP the wrong choice for storage extension. This is no different than data on a DS-3 connection.

Problems here-

1) no practical loss target in the experiment. should have set up a 99% PDR..then we'll see
2) you have done a private network IP experiment---is this real? does this represent what is to be expected in wide scale IP services? IS this what sprint's network will deliver? will you guarantee it? where can I sign up to the only usefull IP service?
3) what is the net amount of equipment this needed to get FC onto optical? FCIP, Router, SONET/DWDM. does this sound like a money saver?
5) how many enterprise IP services can sustain 100MBps? what happens if everone does that?
6) Where is the service demarcation Sprint? What was the SLA? Hey Hitachi---- did you think to ask those questions? you know, a service provider network does not always work like a private network....

Sprint, the problem with this industry is that people(s) like you need to know better. Get your head out of your $sses. This is complete BS.

This reminds me of 'lambdas are free and everywhere' of 2000. I suppose 3 years from now, the marketing will have to face the music on this one too.

Sign In