Larry Youngren, retired leader of IBM’s remote journaling team, recently joined Becky Hjellming, Vision Solutions’ director of product marketing and host of Vision’s PowerTalk podcast, for a discussion of all things remote journaling for IBM i. Larry, a 30 year veteran of IBM, writes, lectures and consults on high availability choices and performance tuning in journal intensive environments, and he shared a wealth of information from his experience.
Larry began by describing the history of the creation of this unique technology. As Larry explained, journaling was created from necessity after a Spanish customer’s disk went belly up and they realized they had to be a better way to “put Humpty Dumpty back together again.” Larry’s team looked at research from IBM California and found a technique for producing a replica copy of each row written to the database, and thus journaling was born to deliver the recovery point objectives (RPOs) that businesses needed.
Yet, challenges restoring systems after crashes continued to plague the team, and it became obvious that the whole machine, not just the database, needed redundancy. So the concept of twin journal receivers, journal receivers that resided on different disk spindles, was created. But even twin journal receivers was not enough when dealing with customers who had suffered through regional disasters, such as a California earthquake. The IBM team determined that they needed a means to get one of those twin journal receivers to a remote site.
So, a group was created whose goal was to create a facility that delivered the needed recovery point objective by enabling remote journal receivers. Their requirements including providing robustness and reliability along with the efficiency required for people to adopt the technology without finding it a burden. And that technology was remote journaling.
Larry provided his insights into the advantages of using remote journaling for high availability, saying. “It assures one important thing. That you get a timely recovery point. Because before you have actually changed the database (let alone on disk), you already have the information ready and are prepared to send it to the target machine. This all happens at the microcode level so you get low overhead and great efficiency.”
Larry also shared four tips for tuning remote journaling bandwidth and his advice on tuning remote journaling performance. One key take away from that part of the conversation was Larry’s Goldilocks Transport Principle: Send not too much, and not too little. Send exactly what you need and not a byte more.
The PowerTalk episode closes with Larry providing his advice on the three factors to consider when choosing an HA solution. First, work with a provider that understands performance tuning in an HA environment and understands transaction integrity. Second, make sure you have clean networks. And third, and ensure you truly understand the bandwidth demands of the approach you select.
For all the details of this fascinating conversation, and to hear Larry’s tips for tuning remote journaling, visit www.visionsolutions.com/PowerTalk today!