Shield Takes Another Stab At Multi Node In HA4i
April 4, 2018 Alex Woodie
Shield Advanced Solutions is putting the finishing touches on a new version of its HA4i high availability software that can replicate data across multiple nodes. The product, which has the working name HA4i Multi-Node, will allow customers to adopt more complex data replication flows, such as one-to-many and many-to-one, which puts it on a more solid footing with other HA products.
IBM i shops have been steadily adopting high availability software over the past few years, studies such as HelpSystems‘ annual IBM i Marketplace Survey and Vision Solutions‘ State of Resilience show. We’re now at a point where about 50 percent of the IBM i installed base has some sort of high availability, either logical replication or hardware-based replications, such as IBM‘s PowerHA.
Most of the recent growth in the HA market has come from the lower end of the IBM i market, which had previously been unable to justify six-figure price tags that HA implementations typically involve. The largest IBM i firms that can least justify downtime implemented HA long ago, in most cases.
Shield Advanced Solutions has been one of the benefactors of this growth with HA4i, which is a budget-friendly HA product that the company says is simple but effective. HelpSystems, which acquired Bugbusters and its RSF-HA product in late 2016, is another.
While Shield started out on the low-end of the HA spectrum, the company is steadily moving up the functionality ladder. A great example of this is the forthcoming release of a new multi-node implementation of the HA4i software. “Many people have asked for a version of our HA4i product which would support multiple replication paths,” Shield president Chris Hird wrote on his company’s blog last month, “but we have been avoiding that until now!”
Hird writes that Shield is close to releasing a new product that allows a multi-node configuration in HA4i. It will have all of the underlying technology of HA4i, in addition to the extra functionality that enables the software to run across a cluster larger than two IBM i server nodes, but it will be a separately licensed product.
The new multi-node product is well into the testing phase, and Hird expects to formally announce the new product sometime before the PowerUp 2018 conference (the new name for COMMON‘s Annual Meeting and Exposition). It will not be an upgrade from HA4i, as the company wants to maintain that product as is.
“We are not going to change HA4i,” Hird writes. “It will remain our product of choice for those who want simple yet effective one-to-one replication for high availability. This [new] product is going to be aimed at the enterprise-level customers who are most likely to require the multi-node capabilities.”
The multi-node feature will allow customers to replicate data from a production system to multiple backup systems. This might include an HA node that can step in and take over processing if the production system goes down, or a disaster recovery (DR) system that the organization relies on to be the last hope in the event of a wider outage.
Many longstanding HA products in the IBM i market offer multi-node features. Some enterprise clients may use multi-node replication to provide a hot backup of the production machine located in the same data center or campus, and then replicate data off the hot backup to a DR machine that’s located hundreds or thousands of miles away, thereby providing geographic separation, which is considered a best practice for enterprise IT shops.
Shield actually offered a multi-node capability in HA4i years ago, but Hird removed it. “Nobody needed it,” Hird told IT Jungle in an interview last year. “All our customers are one-to-one. The majority of customers are going to be one-to-one. Therefore, it was making everybody suffer with a multi-node configuration when they didn’t need it. It became too complex.”
Hird, who is the primary developer for HA4i and also the head of his Ontario, Canada, business, admits that the multi-node feature will add more complexity. However, Hird says he put a lot of effort into making sure the new feature is as simple to use as possible – and has no impact at all on HA4i customers who don’t need multi-node capabilities.
“Users should not need a specialist to come in and configure the product, so it was important to add as much self-awareness and self-help into the product setup and management [as possible],” Hird writes in another blog post. “Lots of new commands and additional tools have been added to help make things simpler such as automated remote journal builds and automated status checking and recovery.”