Vision Sees Positive Trends for HA/DR in Second 'State of Resilience' Report

Published: December 7, 2009

by Alex Woodie

Power Systems users are becoming more stringent over how much time it should take them to recover data following a disaster, according to Vision Solutions' second annual "State of Resilience" report, which the vendor officially unveils tomorrow. That's the good news. The bad news is that organizations are still not prepared to actually recover as quickly as they want, says Vision, which also tried to shed light on IBM i and AIX usage in Power Systems shops as part of the year-long study.

In a perfect world, everybody would have zero tolerance for lost data and downtime as the result of a disaster. But of course, we live in an imperfect world. So, just as individuals can choose from an array of liability insurance programs--each with a related premium--to protect their livelihoods in the event of a disaster on the road or in the home, companies also must make choices about how much they are willing to pay to obtain the level of data protection they deem necessary.

According to Vision's latest report, Power Systems shops want to have stringent DR policies in place, as reflected in their reported recovery point objective (RPOs) and recovery time objectives (RTOs). The problem is, they haven't made the necessary sacrifices in time and money to actually achieve their stated goals. This is reflected in Vision's report, titled "The State of IBM Power Systems Technology and Resilience," which is which based on the responses of more than 4,000 IBM i and AIX professionals to 17 different questionnaires.

For example, take the reported RTOs. Last year, Vision reported that about 40 percent of IBM i shops aimed to have their data recovered within six hours of a disaster or complete server or application failure. This year, Vision says about 57 percent of IBM i shops aim for a six-hour RTO. That's an increase of more than 40 percent in the actual number of IBM i shops who say they will recover their data within six hours.

(The one caveat to this RTO data is the fact that the questions asked in 2008 to 2009 are not identical, a Vision manager said. The company also added another category--recovery of data within an hour--to the 2009 RTO questionnaire. About 24 percent of IBM i shops say they aim to recover within one hour, while 33 percent say it will take from one hour to six hours to recover their data.)

When it comes to the level of accepted data loss, as reflected by the RPO metric, companies are also increasingly bullish about their stated desires, if not their actual capabilities. Among both IBM i and AIX shops (as well as the fraction of respondents that that use both), 40 percent of organizations reported that they have no tolerance for any data loss--that is, their RPO is zero, which is only possible through advanced clustering or failover technologies. That reported percentage is five points higher than last year, according to Vision.

While the improvements in self-reported RPO and RTO goals are good signs, Vision's experience and knowledge of the actual DR and HA processes that organization's have in place indicate that a number of respondents were over-promising their RPO and RTO capabilities, says Chris "CT" Thomas, a marketing programs manager with Vision Solutions, and one of the key author's of the new report.

"Some companies certainly are going to be in for a rude shock in a disaster, and I don't mean the actual disaster that takes out their system, but the disaster of 'Oh my God, we can't get our data back' or 'We're down for five days now,'" Thomas says. "When you look at the difference between their intended RTOs and RPOs, and separately when you glean real experience--when you test your DR program or when you really had to invoke it, what was your real experience--there seems to be a bit of a gap."

Traditionally, tape is the default DR fallback for the majority of customers, and it continues to be, according to Vision's report, which found that 92 percent of IBM i customers rely on tape backup and offsite storage, compared to about 85 percent for AIX shops. About one-third of IBM i and AIX shops use some form of logical replication, clustering, or both, according to the report (with AIX shops preferring clustering and IBM i shops preferring logical replications, per the platforms' differences in storage architectures, Vision says).

Less than 20 percent of IBM i shops use one or more of the remaining DR or HA technologies, including (in descending order of popularity): cross site mirroring (XSM), virtual tape libraries (VTL), Flash Copy, continuous data protection (CDP), Geographic Mirroring, Metro Mirroring, Global Mirroring, and SRDF, EMC's version of IBM's storage-based mirroring solutions for large SAN arrays.

While a minority of companies are adopting higher-end HA and DR products and processes, that should not to take away from the iterative DR improvements that many IBM i and AIX shops of all sizes are making. "The sky is not falling," Thomas says. "Most companies do a pretty darn good job just with tape backup and the tools they have, but there's a lot of room for improvement in the areas we call mini-disasters."

"Mini disasters" are small-scale losses of data, as opposed to natural or manmade disasters that take down entire data centers and business units. Vision sees increased awareness of mini-disasters driving an up-tick in adoption of CDP, which is built into its MIMIX and EchoStream replication and failover solutions for IBM i and AIX. CDP is also available as an add-on to iTera HA and ORION product customers or as a stand-alone product, through a separate offering it calls RecoverNow.

"The increased interest in CDP and increasingly stringent RTOs and RPOs that are quoted are indicative of a mindset among companies that they are becoming more aware that they need to remain available and remain resilient," Thomas says. "They need to be more resilient, and need to be able to recover at a more granular level and more quickly than they did in the past. The question is just how resilient do you need your company to be. How much treasure and time are you willing to spend ensuring that you are exactly that resilient?"

Vision also attempted to gain some insight into "dual citizenship" shops, or the adoption of AIX by predominantly IBM i shops, and vice versa. More than half of the respondents to Vision's survey were IBM i-only shops, while 25 percent were AIX-only shops, and about 21 percent reported being "dual use" shops. These numbers should not be considered a reflection of the actual makeup of the modern Power Systems customer base, Thomas adds.

As would be expected, companies that use both IBM i and AIX tend to be larger in size. Compared to IBM i-only shops, organizations that use both platforms tend to be bigger adopters of all kinds of HA and DR products and technologies.

"We know the profile pretty definitively of small, medium, and large companies in the i community," Thomas says. "We're taking into that the AIX community, but we don't have the same thousands of customers at this point to say we know definitively what the profile is."

Vision will be holding a Webinar tomorrow at 10 a.m. CST to discuss the State of Resilience in Power Systems. The Webinar will feature Ed Vesely, Vision's senior vice president of worldwide marketing, and Mark Shearer, vice president of marketing, communications, and sales support for IBM's Systems and Technology Group. Interested parties can register for the free online event at Vision's Web site at www.visionsolutions.com.

RELATED STORIES

Vision Lays Out HA and DR Options in Well-Written White Paper

Tape Backup Recovery Points Improved With RecoverNow

Midrange Shops Not As Protected from Disaster As They Think, Vision Finds

                     Post this story to del.icio.us
               Post this story to Digg
    Post this story to Slashdot