Startup Looks To Take the Pain Out Of HA Testing
November 3, 2014 Alex Woodie
One of the big issues with IBM i high availability is the lack of testing by the users. The problem is, without regular testing, the chances of a successfully completing a real failover when it counts are slim to none. Last week, a Florida startup called Focal Point Solutions Group unveiled a hosted high availability offering called DR FlashCopy that uses an innovative approach to allowing customers test their HA environments at their leisure.
“I hate to say it but the majority of high availability customers haven’t done a switch in several years,” says Ron Venzin, the CEO of Focal Point Solutions and a veteran in the high availability business. “A lot of customers are petrified to take a look at it, other than to see if it’s replicating. They just hope it works when they need it.”
Crossing your fingers is a natural reaction to things outside your control, like earthquakes, tornados, hurricanes, fires, floods, locusts, and zombies. But it boggles the mind that a company would trust its fate to chance after investing $50,000 to $100,000 in a high availability system that should allow it to stay in business when Mother Nature has a bad day.
While the “set it and forget” mentality is widespread, it’s incompatible with logical replication solutions, no matter what vendors they are from. It’s important to keep HA implementations up-to-date with changes made to databases and applications. Data groups need to be updated as a regular course of business, and object groups need updating when the application changes and other changes are made to the system.
Without this regular maintenance, a company has no guarantee that the secondary IBM i instance will be able to support the business during a disaster. And without regular testing, a company likely won’t see the problems. It’s difficult for many IBM i shops to get the outage windows necessary to adequately test the setup, because many of these systems run around the clock. After all, that’s why they brought in HA in the first place!
It’s a nasty little conundrum, but the folks at Focal Point Solutions Group may have come up with an innovative solution to the problem.
Focal Point Solutions Group was founded in June by Venzin and several other IBM i technology veterans, including Garrett Schut and Pete Elliot. The company is a managed service provider (MSP) with a focus on hosting IBM i environments, including both production and replicated instances. It is partners with Vision Solutions and Baseline, and uses the data center resources provided by zColo.
As an MSP, Focal Point Solutions implements and manages Vision’s HA products running in a private cloud environment. It has expertise with MIMIX and iTera, while its data center agreements give it access to IBM i servers that its customers can use, either for production or backup workloads. Up to this point, the company sounds a lot like other MSPs.
But DR FlashCopy introduces an interesting twist on this established MSP story: Instead of just hosting HA on plain vanilla IBM i gear, it uses external storage based on IBM Storwize arrays. One of the benefits of using Storwize gear is it allows customers to use IBM’s FlashCopy technology, which makes full copies of an IBM i partition very quickly–in a matter of minutes, in most cases.
Focal Point Solutions uses FlashCopy to create a full copy of its customer’s environment, which they can then use to test at their convenience. The customer can take as much time to run tests on the copied environment, while the Vision-based HA solution continues to replicate transactions. This approach eliminates the need to bring a production server down for an extended amount of time in order to test the HA setup.
“We can do a switch test at any time,” Venzin tells IT Jungle. “They can make a FlashCopy on Tuesday, test it for three days, and they never stop their replication. They never touch their source or their target information, because we flashed it to another environment for them to test it on. Because the flash occurs so quickly, we’re able to test the environment, destroy it, reflash it, and start testing it over.”
The idea is to tear down the boundaries that have prevented them from testing their HA environments, and to give them the type of continuous availability that the biggest enterprises enjoy. “It gives them the flexibility and cost savings, because now they’re not sending people to remote sites on weekends. Now you’re not working at 3 o’clock in the morning and you’re not worried about whether you can get this test done in your four-hour outage window. You don’t have to worry about an outage window anymore.”
FlashCopy as a Service
It is true that an IBM i shop could get their hands on their own Storwize array and start using FlashCopy to speed up the test cycle. Storwize systems, particularly the V3700 and V5000 models, are starting to show up in midsize businesses. But this sort of technology is still too advanced for the typical average small or medium-sized business (SMB), Schut says.
“There’s a lot of steps involved in doing things on the HMC and steps on the V7000, which is typically a skill set that does not exist in an SMB environment,” Schut says. “So we’re able to provide them with an enterprise-level solution, with testing capability that you’d never be able to do in an HA environment.”
In addition to HA testing, the DR FlashCopy offering can be used to speed backups and reduce the vulnerability that users have during backups. Instead of running a backup off the replicated copy–which typically requires the journal apply process to be temporarily halted–the customer can use FlashCopy to create a full “golden” copy of the data. The offering can also be used to test other applications, such as ERP upgrades. For example, if a JD Edwards customer wanted to see how an upgrade was going to affect the application, it could spin up FlashCopy and test the system to its heart’s content in a safe sandboxed environment.
There are some caveats to FlashCopy. For starters it only works with IBM i version 7.1 or higher. And while it copies partitions very fast, it still needs five to 10 minutes to work. That means you probably don’t want to create a copy during the busiest times of day. “You’re going to lose anything in flight, obviously,” Venzin says. “Most customers will flash early in the morning or late in the day or at lunchtime. We would recommend flashing a system when there is less activity on the system for you to have the cleanest availability test.”
Focal Point Solutions isn’t the first MSP providing hosted HA as a service. But with its innovative use of IBM’s FlashCopy technology, it may be a trendsetter nonetheless.