HP to partner with Cisco, Oracle on high availability
Plans to integrate management software with ClusterView HA software
HP High Availability Product Manager Bill Layton says that the partnerships with Cisco will extend the work HP has been doing on its Mission Critical Server Suite, announced last year, to incorporate more parts of the network in HP's high availability product line.
Layton says that offering high availability at the operating system and hardware level is often not enough for real-world applications. "Who cares if your system is up if your clients don't have connection to the database?" he asks.
Discussions for similar partnerships are also underway with potential partners, such as SAP, Baan, and PeopleSoft, according to Layton.
Right now, HP is vague on exactly what services it intends to provide with the Cisco and Oracle partnerships, but there are some hints. HP Product Manager Joaquin Ruiz says that HP's ClusterView monitoring software will be integrated with upcoming self-managed databases from Oracle so that a database could be monitored and tuned from the HP console. HP is also promising "technology" that will reduce Oracle Parallel Server recovery time on HP-UX.
Cisco is expected to integrate a number of its tools into the HP OpenView environment, including a network disaster impact modeling tool and software to re-route clients in the event of router failure.
The plan is for customers to have a single point of contact within HP for application, router, and system service when products become available later this year, according to Ruiz.
The road to fault tolerance
Though companies like HP and Sun have incorporated high availability services in specific applications and network configurations on an ad hoc basis in the past, Layton claims that "this is the first time that Cisco and Oracle have stepped up with a major vendor like HP to collaborate on high availability." He says by making client-to-database availability part of the product line, HP will both strengthen its OS and reduce the cost of these services.
Sun, for its part, guarantees application availability on a case by case basis, says Solaris Clustering Product Manager Rakesh Singh. He says Sun hopes to "at some point in time" be able to offer high availability guarantees for individual applications, but, he adds, "I don't think we're on the verge of making that announcement."
Harvey Hindin, the director of D.H. Brown's high availability and clusters group says offering mainframe-type reliability is becoming more important in the open systems world. "It used to be that 7x24 [availability] was important to the finance and telecommunications industries, where downtime was expensive," he says. "Now, more and more industries have to do that. I think HP is right on the mark in realizing this."
HP says that over the next year it will clarify the specifics of its partnerships with vendors like Cisco and Oracle -- how it will organize sales and service, what kind of software tuning it will do, and how, precisely it will guarantee uptime. Hindin cautions that it remains to be seen whether or not HP can actually implement these true high availability services from end to end. "This is nitty gritty engineering, and it goes past press releases," he says. "The genius is in the implementation."
D.H. Brown's Hindin predicts that not only will Sun and IBM have to take notice of HP's moves into high availability, but that that vendors in the fault tolerant community -- the Tandems and Stratuses of the world -- will also have to keep their eyes on HP or risk losing business.
In early spring, HP will deliver an enhanced version of ClusterView that will integrate with HP management and administration software like its Event Monitoring Service SNMP monitoring software and its Process Resource Manager CPU management product. Ruiz says that administrators will be able to manage things like disks, file systems, and network interface cards with the enhanced version.
By the end of 1998 HP says administrators will be able to define application packages (for example, a software patch downloaded from the Internet) and then drag and drop them into many different nodes on the network.
Also by year's end, HP plans to add kernel hardening technology developed with Hitachi, based on the latter's Exception Infrastructure technology. This will be designed to reduce system panics and hangs by enabling the operating system to automatically backtrack to its last good state in the event of trouble.
In the post 1998 timeframe, HP's high availability roadmap calls for:
If you have technical problems with this magazine, contact firstname.lastname@example.org