In the heart of his Montréal command centre, CN’s computer systems officer, Brian MacKenzie, sees everything.
MacKenzie’s team supports the rail company’s data centre operations, known as the Information Technology Command Centre. It’s a 7×24 operation, and the front line when it comes to alerts
about any application, database or operating system.
“”We’ve also brought in the networking management centre portion,”” says MacKenzie, “”which is a lot of the fibre and the wide-area network across CN’s enterprise.”” The team manages the tools for those operations, configures and implements accordingly.
“”We cover data storage management, monitoring and automation, and scheduling,”” McKenzie says. “”We’re also involved in disaster recovery.””
To make this job easier, CN selected Hewlett-Packard’s Openview software to monitor, report and provide centralized systems management for its distributed, heterogeneous IT infrastructure.
“”It’s a centrally managed operation in Montréal,”” says MacKenzie. “”We have 32 field sites where we are monitoring servers and backups and so on. They range from Moncton to Prince George and as far south as Memphis, Tennessee.””
From this central point of command, Openview will automatically uncover bottlenecks and initiate actions. It’s a big picture to keep an eye on.
MacKenzie, whose background is in mainframe technology, says CN started an automation task force as far back as 1988 for the mainframe environment. “”It built from there. We knew we wanted to put in the same practices as we had with the mainframe.””
In early 1999, CN began implementing SAP. “”The decision at the time was not to take any risks with SAP,”” says MacKenzie, so CN opted to roll it out on an HP UX servers. “”Once the phase one of SAP was complete, then Openview was handed over to our team to manage.””
MacKenzie took it upon himself to investigate the advantages of Openview sitting on an AIX platform and monitoring the various databases. “”We started with AIX, we started with different proprietary applications,”” he says. “”Then we went over to NT, then we started to branch out to the other field operations. It’s been an ongoing project ever since.””
The challenge of deploying Openview, says MacKenzie, is getting all the players together to work on it — the DBAs, the systems administrators.
He says CN’s recent Windows 2000 deployment to replace Banyan Vines was a big step — it was change implemented across the company’s 32 sites. Once the environment was stabilized, Mac-Kenzie says the decision was made to use Openview for monitoring, making the software the standard for watching CN systems.
“”When we have a hardware failure or any type of alert coming in, it comes to the command centre here in Montréal,”” says MacKenzie. Staff in the command centre will try to fix the problem or bring the appropriate people to solve the problem — and a problem can have a significant real-time impact on CN’s business.
“”The servers we monitor in Montréal and also in the yards affect
the operations,”” says MacKenzie. “”There are servers that are managing our automated gate systems for our intermodal operations. That’s primarily for when trucks come into the yards and they go through a biometric-type system. This way the system knows exactly which truck this is, what container it is and where it’s supposed to go.””
Up to 4,000 trucks can enter a CN yard in one day.
“”There can be no delays of that system,”” says MacKenzie.
Rejean Lafontaine, senior systems programmer at CN, says the strategy for deployment has been to start with a basic task or process the organization wants to monitor. Some agents are specifically programmed for their task; others use the default features.
“”The package comes with a lot of already pre-coded monitors,”” he says.
Lafontaine has worked with automation tools for 15 years, and says automation used to be a big deal — there was a lot of configuration needed. Now, he says, everything comes right out of the box; features can be turned on or off.
While automation tools do a lot more than they used to, pure autonomic computing will take some time, says Lafontaine.
“”It does a lot for you, especially to collect the data,”” he says. “”With current “”self -healing”” systems, he says, easy problems can be fixed. “”As soon as it gets a bit complex, you have to intervene.””
CN is a pilot customer for HP Software Self-Healing Services for Openview and will be testing technology that combines an embedded software engine with integrated support services.