iTera Says Out-of-Synch Conditions are a Thing of the Past
October 25, 2005 Alex Woodie
Perhaps the biggest challenge of maintaining an OS/400 high availability implementation is dealing with out-of-synch conditions. When a file or an object on the backup system doesn’t match up with the corresponding file or object on the primary system, an OS/400 shop risks losing data when a failover is performed. iTera says it has created new technology with the latest release of its Echo2 high availability software that makes worrying about such conditions a thing of the past.
Last week, iTera unveiled Echo2 version 4.2, which the company started selling October 1. This release contains several “autonomic” functions, which is to say iTera has worked to automate some of the everyday tasks that typically take up operators’ time.
One of those tasks has historically been running audit checks to make sure the backup system is in synch with the primary system, says iTera President Dan NeVille. iTera calls these circular redundancy checks (CRCs), and the company provides a tool that goes out and performs CRC checks across every record, file, object, and library to make sure they match.
Running CRC checks and re-synchronizing two boxes or partitions is not a big deal when you have a very fast optical link connecting them, NeVille says. However, when large files and wide distances are involved, the current method of detection and repair, which involves sending updates over the network–or worse, by tape and mail courier–is less than ideal. In fact, when it takes a day or two just to make sure the two systems are in synch, it takes away the whole point of having high availability in the first place.
Dealing with out-of-synch conditions is a fact of life in high availability environments, NeVille says. “Our product stays in synch pretty well,” he says. “But things will still get out of synch. Then you’re at risk until you can get that back in synch.”
‘More Important than Remote Journaling’
iTera’s answer to the out-of-synch problem and current detection and repair methods is something called Concurrent Audit Concurrent Heal, or CACHe2, which it debuted in Echo2 version 4.2. This new technology continually looks for out-of-synch conditions at the record level, and automatically fixes them, without operator intervention, iTera says. The new CACHe2 technology does not run up the processor bill, or have any other affect on performance, NeVille says.
NeVille says CACHe2 solves one of the vexing problems of high availability. “The problem with high availability solution is they get out of synch, and it’s tough to keep them in synch. It’s always been an art form to keep systems in synch,” he says. “With this record level heal, we look at the difference with the records, and we fix it. Nobody has ever been had the capability to check in real-time, and also repair in real time.”
CACHe2 was developed more than a year ago, and has been in use for six months at select iTera shops. iTera applied for a patent for CACHe2 with the U.S. Patent Office 18 months ago, says NeVille, who credits his director of research and development, Jeff Asherman, with creating “the greatest thing to ever happen to high availability.”
NeVille minces no words in describing how important he thinks this new capability to detect and correct record-level discrepancies in real time will be to the high availability business. “We believe this is more important than remote journaling,” he says. “It’s miraculous.”
In fact, CACHe2 will have such an impact on high availability, NeVille predicts, that it could obviate the need for the new cross site mirroring (XSM) capabilities IBM has baked into recent releases of OS/400. “We believe that if IBM had any inkling of this, they never would have tried to do switch disk. The Large User Group asked for that because they could not make high availability work.”
iTera has built additional autonomic functions into Echo2 version 4.2, in addition to CACHe2. One of these is E2 Apply Accelerator, which is designed to reduce latency, which is the time lag between when a transaction occurs on the primary machine and when it’s applied on the backup machine–another common area of concern in high availability environments. The technology automatically detects if the apply process is falling behind due to increased traffic, and if so, automatically engages processes to quickly eliminate latency, the company says. NeVille says this feature will provide the most benefit to very large shops processing millions of transactions a day.
The new release of Echo2 also sports a new E2KG feature, a scheduler of sorts designed to ensure that audit tests are run on a regular basis. Like all high availability software vendors, iTera recommends that users periodically check their system to ensure it can successfully handle a switchover if called upon. Being ready to carry out a failover with little notice is the whole point of running high availability software, but unfortunately, running the checks that tell you if you’re actually ready to flip the switch is an area that many OS/400 shops are slipshod in checking, no matter how much education vendors apply to the market. By automating these checks with E2KG (which refers to an EKG heart monitor), iTera is aiming to keep users abreast of replicated environments’ little “gotchas,” while reducing an operator’s need to baby sit the system at the same time.
Finally, Echo2 4.2 also boasts a new transport method for sending changes made to spool files. iTera uses remote journaling as the foundation of its product. But because remote journaling only supports the transmission of changes to data, data areas, data queues, and Integrated File System (IFS) files, iTera had to look somewhere else for ferrying across changes to things like spool files, programs, and directory entries.
NeVille says the new transport method resembles remote journaling. Actually, he says it is remote journaling. “We have created our own remote journaling,” says NeVille, adding that if IBM ever built support for spool files, programs, and directory entries into remote journaling, then iTera wouldn’t have to use this alternative remote journaling.
It’s been about a year-and-a-half since iTera updated this product; see “iTera Beefs Up iSeries HA App with MQ, Auditing Features” for our report on the last release. If the company’s 2004 accomplishments are any indication, Echo2 4.1 was a big seller for the Salt Lake City company, which has only 65 employees but 1,200 Echo2 customers, and is currently expanding into Europe, Asia, and Latin America.
NeVille says iTera is aiming to build more autonomics into its products. “We’re trying to make our software able to run itself. We are real close,” he says. “When we first came out, customers would spend 30 minutes a day managing high availability. Last year, it went down to 10 to 15 minutes a day. Now we think we’re down to one to two minutes a day spent managing high availability.”
Echo2 4.2 is generally available. The product supports OS/400 V5R1 through V5R3 (although the CACHe2 function is not supported on V5R1). Pricing is tier-based and ranges from $10,000 to $80,000. For more information, visit www.iterainc.com.