Dos and Don'ts of DR with Richard D
Published: August 14, 2012
by Alex Woodie
Richard Dolewski, a disaster recovery (DR) expert with Velocity Technology Services, has seen things in the DR world that will make you cringe, such as the Fortune 500 company without a DR plan or recoverable systems. With extreme weather events on the rise and companies' appetites for downtime on the decline, now is a great time to re-evaluate your strategy for keeping your IBM i apps running during a disaster. Dolewski recently shared some dos and don'ts on the topic with IT Jungle.
DO--Make sure that you have backed up all the necessary IBM i objects and other infrastructure components that are necessary to rebuild your IBM i environment, not just the data.
I always ask people in my sessions, 'Do you back up the data?' And 100 percent say they do. Well, over 40 percent of them cannot recover their primary and critical interfacing systems. The big thing is, it's the infrastructure components that they haven't protected properly. Many backup policies are incomplete, even if they're running it natively on the platform of choice, and using something proven like BRMS (Backup Recovery and Media Services). Examine your BRMS recovery reports and you will discover just how complicated a recovery awaits you.
I have seen companies that have backup policies of yesterday's purchase orders, but lack application or operating system policies. They can restore the purchase orders, but there's nothing to recover them into. We're evolving so fast, that backup technologies and policies are often out of date with the current implementation of the infrastructure.
DON'T--Automatically assume that storing your IBM i backups on a non-IBM i platform--as many cloud-based data vaulting solutions do--is a proven commodity.
My dear friends at IBM would support me on that. It's not a certified process. There's a trend out there to take i data and incorporate the use of cloud privately or even publicly, to back up to and then ultimately recover from other "open" file systems. The fact is, you potentially change the file structures and could potentially alter the data or layout of the data by going from a proven technology, transferring data to another, and then trying to restore it back to its original state.
I think we're overcomplicating something that's been a very simplified process from IBM. It's proven, it's rich, and it works. I don't think I would bet my business going from an i over to a Linux- or Windows-based file structure and then going back to i and hoping that it works 100 percent every time. I just don't know if that's a repeatable process. The IBM i is a unique platform, and I think using non-IBM i solutions, outside of what this thing was built on, is a dangerous proposition to my business.
DO–Be confident and positive about getting company executives involved in DR strategies and decision making.
Look at all the companies that went down in late June with [the Derecho storms that hit the East Coast]. A tree falls down, it takes out a generator. But if it takes out my data center and I can't recover it and I'm not up until Wednesday of next week, then it's no longer funny. Imagine not being able to sell product for four or five days. There goes my stock price, there goes everything my company strives for.
We're finding a stronger alignment of business and IT than ever. Go back five years, and people would analyze this thing called DR, hire some consultant, talk to finance guys, and sit there and do all sorts of wonderful data models that would say, 'Here's what we need to do should we have a disaster, but the percentages say we won't, so we're not going to do anything." Today, data protection is of the utmost importanance. Unplanned interruption in business, whether it be human error or technology, is just unacceptable these days.
DON'T--Update your business systems with the latest technology without also revaluating your existing backup strategy and making updates if necessary.
Businesses have to align their data protection schemes, operationally and strategically with their disaster recovery goals. The money and the emphasis lately has been for companies to incorporate new cloud or new POWER technology. Server virtualization, whether it be PowerVM or VMware, has hit mainstream, but unfortunately we haven't kept up with today's resiliency solutions to make our infrastructure more available. We haven't kept pace by improving the backup and recovery side of the equation.
We need to be more efficient, and innovative, in the adoption of new technologies, whether it be software replication, POWERHA, live partition mobility, or whatever. We've done a good job keeping the data center up to date and making it resilient, but we are shifting away far too slowly from slow foundational DR technologies, like tape backup. Can we still afford to wait to recover the business when there are recovery time and data protection techniques exists today that can bridge the gap?
DO--Consider native IBM i data vaulting, or continuous data protection (CDP), as a DR solution to bridge the gap between tape and HA.
How do you take care of these mini-disasters, such as power outages and network outages? We have technologies to recover from a complete disaster, but now you need to evolve those technologies to deliver systems availability or resiliency, whether it's replication or clustering based. These are all forms of keeping data and systems resiliency in place so that you've got real time protection of data as it continues to evolve. Data from analysts clearly shows there's a resiliency need out there. You've got tape technology on the one side, and on the other side you've got HA like Vision Solutions or IBM's PowerHA. But in the middle, there isn't a heck of a lot that exists today.
What we've done at Velocity with the use of replication software and internal IP is created fully managed data vaulting on the transactional level for clients, so that they don't have to go all the way to HA, where I need two like systems, and I can't afford that. And yet I don't want to suffer the data losses as I would with tape. We deliver an IBM i solution on IBM i. It comes at a good price point. If you were to spend $2 for a tape, commodity-based recovery, and spend $10 for HA, this comes in at $5 or $6. Maybe as much as 70 percent of Velocity customers ask for this over tape and HA.
DON'T--Automatically assume that high availability is too expensive for your company.
The price point for high availability software is very attractive right now. It's probably a third to a half of what it was three or four years ago. We're getting a measurable trend that people are adopting it and can afford it. Remember: Tape will still be the firewall, the stopgap for catastrophic events. But we're finding that the cost of disk storage, the cost of replication software from all vendors, is making this an economic play. And business is driving that too, because they want real-time protection of data.
DO--Research the ISVs behind a high availability software product before implementing it.
Some of the [less expensive HA solutions] are compelling. But I would worry about some of the one- and two-man operations that have hit the market. You're betting your business on these people to continue to develop their product and evolve it as IBM OSes evolve, or as application software that runs on the i evolves. I like competition. I think it's good for everybody. I think there are some reasonable competitors out there that do have some good products. I've played with Maxava. I like it a lot. They've done some mainstream things to evolve the industry, and others have followed. I, personally, am a little old school. I wouldn't bank my business on a vendor that's a one- or two-person operation, where their level of commitment and financial backing would concern me. I would be cautious on some of those that are here today and gone tomorrow.
DON'T--Sign a contract for business continuity services for your IBM i infrastructure without looking closely at the fine print.
The subscription rate, to me, is the number one concern. Another thing to consider is access time. What happens when I have a disaster? Is there a disaster declaration fee? What about FEMA region separation? If you're in the Florida panhandle or anywhere in neighboring states, what are the odds of several clients in several states declaring a disaster when Hurricane XYZ comes up the coast? It's pretty high. The subscription of customers to the box is high from DR providers.
And who decides if it's a disaster? A lot of vendors do not let you declare a disaster until the governor declares a disaster or a federal state of emergency has been declared, or your data center is under six feet of water. And then what happens when you get there? Many vendors charge you a very high daily usage fee. There's nothing like paying a monthly fee, paying a declaration fee, and then getting the full rack rate like you would in a hotel to run your business there. That’s a lot of extras on top of that monthly fee.
Post this story to del.icio.us
Post this story to Digg
Post this story to Slashdot