The next time you sit down in the evening to solve a jigsaw puzzle, think for a minute how nice it would be if you had someone looking over your shoulder spotting the missing pieces for you. Granted, it would take away part of the fun, but you would probably have the table clear again in time for dinner.
Within the data centre, those responsible for IT operations are forced to solve exceedingly complex puzzles every day as they struggle to keep their IT environments up and running, as well as working closely with all levels of business to identify, implement and support new technologies and applications in order to improve service delivery and to ensure optimum productivity.
Unfortunately, this means that new pieces are daily being added to the puzzle. “Operations teams today can no longer simply look at the IT infrastructure in terms of routers or database servers or web servers, they have to be able to look at it in terms of the business processes that run across them,” says Jonathan Haworth, a product marketing manager for HP Operations Centre software based in Birmingham. “By doing so, you become able to make informed decisions about where you invest your resources.”
Take, for example, two event notifications arriving at the same time, one about a paper jam in a printer in your warehouse and the other reporting a hard disk running out of space for transactional information on an Oracle database. Common sense says fix the Oracle database first. However, taking a business service dependency view, it might turn out that the transaction system is out of service hours right now, but trucks may be backing up all the way to the highway because their drivers lack their loading documents. The really smart thing to do, it turns out, may be to fix the printer first.
In the past, operators had to make these decisions on their own. “That’s tricky because, while discovering technology components in our IT environments is relatively easy, the business services running on them are usually much harder to understand,“ Haworth says. Besides, defining the dependencies between the business services and the IT infrastructure has typically been a manual task.
Also, operations departments, often seen by management as bottomless pits into which money disappears (operations can easily consume half or more of the overall IT budget), are under pressure to cut manpower costs.
Three answers, Haworth believes, hold the key to the solution: focus on things that matter to the business, avoid doing anything twice, and automate everything you can.
This is, of course, easier said than done. “The problem with that is that unless you exercise a very high degree of rigour and control in the way you add infrastructure and maintain those business service views, it very quickly becomes inaccurate and stale, and as soon as that happens those views are no use to you at all,” Haworth maintains.
By automatically building up graphical views of their infrastructure and business services, IT operations can take most of the guesswork out of incident handling, Haworth believes. “Thanks to the Mercury acquisition, HP now has a very nice discovery technology that doesn’t just look at the facts, like there’s an Oracle database out there, but also tells you: look, this particular machine is talking to that machine over there, and it’s using this set of ports to do that, so it’s probably an Oracle client talking to an Oracle server.” By taking that discovery information and making it available not only to the business availability applications, but also to operations applications like HP’s Operations Centre software, operators can create a much cleaner and more accurate and automatically maintained view of the relationship between infrastructure and the business services. “That’s a big deal for customers,” Haworth says, “because it helps avoid the effort of manually maintaining those business services. It’s a great example of how automation technologies are going to enable the IT organization to make sure that they are really investing their people’s time in the things that matter most to the business.”
Using HP’s recently launched Operations Manager Dependency Mapping Automation product, data centres can also widen the uses that they make of the data from their discovery processes. Many routine maintenance jobs such as updating security hotfixes on an Exchange server happen automatically today, but the majority of failures – some analysts put the figure at 60 to 80 percent – that openly affects the availability and performance of business services are caused by unauthorised or unplanned changes, Haworth says. “So the ability to automatically discover and report changes to a component, as well as to the surrounding infrastructure, may mean the difference between a prolonged and painful incident and a quick fix,” he believes. Thanks to a set of “smart plug-ins” and solution templates for its Operations Centre software, and by connecting incidents detected by these components to discovered changes, HP can actually help operators automatically find the right solution to a given problem. “If you’re thinking in terms of ROI, streamlining incident analysis, by providing all the relevant information automatically, can dramatically shorten the time from detection to repair.”
So why not simply automate the entire operations department? Because even with scripts and tools there still remain many pieces of the puzzle that must be put in place by hand. While automating simple actions such as finding and deleting temporary files to free up disk space, or restarting a service, are possible today, tackling more complex operator actions with the help of scripts and automated tools remains risky and trouble prone.
However, help may well be on the way. With the acquisition of Opsware (see article Factoring out the human factor), HP has positioned itself to extend automation far beyond what is possible today, Haworth explains. “There is already some integration between the Opsware technologies and some of the Operations Manager products, but we are going to be starting to work on driving those forward and making much more use of them in the Operations Centre context. We see that as being a very powerful way of reducing the amount of manual activity the operations people have to engage in, and of building much more intelligent automation into the operational management solutions.”
So instead of having to puzzle over the bits and pieces of their systems, data centre managers may soon have automated tools to do most of the hard work for them – hopefully allowing them to get home in time to finish that jigsaw puzzle before dinner.
|
Printable version
