NEW YORK – According to users at Storage Decisions, the biggest obstacle to creating a truly effective disaster recovery (DR) strategy isn't technology, logistics, or even finances -- it's getting the support of their organization in creating a realistic plan.
During a session entitled "Ten reasons your DR plan won't work," Glasshouse Technologies Inc. chief technology office Jim Demoulakis placed "Business and IT are not in sync" at the top of his list.
Demoulakis said 71% of companies only involve IT staff in their DR strategy, and only 38% of companies have an integrated business continuity and DR plan.
"What happens is a gap between expectations and reality," he said.
"The biggest challenge is the mentality of some of the people I work with," said Hansram Ramrup, IT director for the Board of Commissioners in Broward County, Fla. "They've been in government for years, and sometimes their mindset can be three or four generations back."
"DR is more about a mentality than anything else," said Dave Naessens, manager of IT for Keihin Carolina System Technology Inc. "IT has to sell itself, but IT engineers don't tend to speak the same language as the business side."
Meanwhile, according to Steve Duplessie, analyst with the Enterprise Strategy Group, recent events have shown that a lack of well-planned DR is an urgent problem. "We almost can't afford not to start thinking about this. No matter who you are or where you live, we've seen that you've got to have lots of miles between your data sets."
So how can IT professionals bridge the gap and make a more effective, realistic DR plan? There is no quick fix, but users had some suggestions.
Keep it concrete
"My director has a pretty open mind, and a technology background," Ramrup said. "Instead of going into his office with a piece of paper, I try to set up a demo, sometimes with a vendor's help, so he can see for himself how a product will actually work. Instead of a salesman, I'll try to get a vendor to send in a technician to come in and do a mockup of their product."
Ramrup said he intended to spearhead an effort to make his DR tests less "staged," in order to show his end users the real effect of their current DR plan.
"Right now, our homegrown plan for recovering Exchange is to put empty mailboxes in front of users and allow them to send and receive messages immediately," he said. "We're totally dependent on tape for backup, so in the best case scenario, it'll take 24 to 48 hours to inject the e-mails back into their mailboxes. In reality, it'll probably take more like 48 to 72 hours, depending on the restore time for the tapes.
"For a government office that does everything through e-mail, that's a huge loss."
"It's important for users to make a business case, with cost-risk numbers to back it up, for the infrastructure they want to see in place," said Pierre Dorion, senior business continuity consultant for Mainland Information Systems Inc. "Many times, when I'm brought in as a consultant, I'm only telling managers what IT's already been telling them -- they just haven't heard it explained in a way that makes things clear to them."
Start a grassroots effort
"Rather than going to my board of directors, I started with lower level coordinators and management, asking what IT could do to better support them," Naessen said. "Whenever middle managers get promoted, I make sure to work closely with their replacements. When I got the ear of people in the lower tiers of the organization, the upper management ended up giving me more funding."
"It's important for IT to find a champion," said Ronald Durbin, director of industry research group StorageNetworking.org. "Find the most articulate person in your department, or someone internally in sales or another business department, who has an affinity with management and can be your representative."
Some IT engineers may even want to bring in their own consultant to make a case to management. "It's always better off bringing someone in than having your boss bring him in," Durbin said.
If all else fails…
Even the best sales pitch sometimes won't work. "Unfortunately," said Alan Bodnar, who requested his company not be named, "it seems like the only way it happens, at least quickly or from scratch, is if you actually go through a major disaster."
There are technologies that can allow storage architects to accomplish more with less. According to Christopher Hill, assistant director of IT at New York law firm Thacher, Proffitt & Wood LLP, virtualizing his servers with VMWare has reduced the time it takes to get his secondary site in New Jersey up and running, as well as the complexity of what it takes to power it up.
"We're not using domain name systems tricks to get it to bring data live. VMWare tells the host in New Jersey where to look for a LUN on the virtual disk, one script registers it, a second powers the secondary server on, and we're finished in terms of going live."
In addition, Hill said virtualizing his servers has allowed him to put more applications into fewer boxes, saving him not only on server hardware but connective hardware as well.
"We were about to buy another 64-port switch," he said, "But by taking out 15 servers in my New York facility, I don't have to buy it now. I just saved $40,000."
If introducing across-the-board measures is also impracticable, Bodnar suggested taking DR implementation piece by piece.
"Test software on a pilot program to see what works. Get one application solidly covered in terms of DR -- and make sure you think in terms of application. Ten servers can be perfectly mirrored, but that doesn't mean anything if the 11th goes down. Think about the front end, what the demand will be, how end users access it."