Part of your job as a server administrator is tracking down stuff that doesn’t work – and then taking steps to fix it. Unfortunately, that job sounds a whole lot easier on paper than it is in practice. Most often, it won’t be entirely clear what’s gone wrong with a server – and you’ll end up pulling out your hair as you try to figure out how you can get things back to normal. It doesn’t help when, as you try to manage your server, you’re faced with a bunch of real-world roadblocks, as well.
Don’t worry – we’re here to help.
No matter what career you’re involved in, you’re really only as good as what you know. Server administration is no exception to that rule – if you’re aware of the roadblocks you’re likely to face, then you’ll find it a whole lot easier to do your job when the time comes to address a malfunctioning server. That’s why today, we’re going to go over a few of the biggest challenges server administrators are forced to deal with – and how you can overcome them.
Managing Memory Pressure
We’ll start with an SQL-specific issue, courtesy of SQL Sentry’s Jason Hall. Perhaps the most frequently-encountered issue with SQL (And one of the most difficult to diagnose) involves something known as memory pressure. Not only that, writes Hall, it’s all-too-often misdiagnosed as an issue with disk performance.
This is, he explains, tied to two incredibly misleading symptoms usually associated with memory pressure: “higher-than-normal latency across the disk subsystem,” and “abnormally high waits related to Disk activity.” If you’re not seeing a complete picture of your server’s performance metrics, then you’re probably going to mess something up when you set out to fix a problem. So…how can you avoid doing that?
“Typically, what I’ll see along with the waits and latency is a PLE (Page Lifetime Expectancy) that is fairly low for the server,” says Hall. “Another consideration is NUMA. The way the PLE counter is calculated can cause this value alone to be very misleading when multiple NUMA nodes are evolved…I also usually see consistently higher lazy writer activity, and SQL Server page faults; sometimes, I’ll see what I call buffer tearing, or an abnormally large plan cache; all of these things together spell memory pressure.”
As to how you can deal with it? Well, according to Hall, the first step is to paint for yourself a complete picture of your server – don’t just pay attention to one metric or detail. Only with a complete understanding of what’s going wrong with your server can you start to deal with memory pressure – external or internal.
Once you know memory pressure is the cause of your server’s poor performance – and not faulty hardware – you’ve a few options. First, you can reconfigure your server’s memory – how it provisions memory to applications and users, how it manages memory overages, et-cetera. You could also consider using alternative applications on your server – software with a lower memory footprint. Finally, you may simply need to order new hardware, or upgrade the stuff you’ve already got.
The Provisioning Of Resources
Our next entry is born entirely of human error. When a business switches over to virtualization, it’s presumably for a few different reasons – to cut costs, perhaps, or to allow for more efficient provisioning of applications. Unfortunately, this can often lead to an organization either spending too much on resources or too little.
Thankfully, this one’s extremely easy to address – as with memory pressure, it’s all a matter of proper monitoring and bookkeeping. Simply advise those in both your organization and department to pay careful attention to their resource usage over the weeks leading up to the switch. Monitor exactly how much memory everyone’s using, then do a bit of research on how much memory each app you’ll use in your virtualized environment will require. From there, it’s a simply matter of extrapolation.
Security, Control, And The Consumerization Of IT
I’m going to go out on a limb here and assume that, as an IT professional, you’re already familiar with the buzz phrase “The Consumerization of IT.” And if you aren’t, you should be. In essence, consumerization is a blanket term used to apply to the blending of personal technology with business technology in the workplace – both software and hardware.
There are a few trends associated with this that you need to keep in mind:
- Because of the ease with which they use their personal smartphones and tablets, employees expect that business solutions be just as intuitive and easily-deployed. If an IT department fails to meet this expectation, they’ll seek alternatives, all of which are potentially less-secure than those provided by the company. It falls to IT professionals and application developers to create for their users a unified, appealing experience.
- Due to the wide variety of devices being used by staff at an organization, it becomes effectively impossible to manage credentials and compliance on a device-by-device basis. Instead, IT professionals must shift to managing identity and credentials.
- As a result of the cloud, IT’s role in deployment is much less pronounced than it used to be. Professionals – including administrators – will be expected to shift into more of an advisory role, while still maintaining an enterprise’s infrastructure. Basically, you’re going to need to familiarize yourself with how cloud computing functions.
Honorable Mention: Managing Troublesome Users (And A Bad Reputation)
There’s really no simple or diplomatic way to say this, so I’m just going to be blunt: every business has at least a few users in its ranks who are really, really stupid. They’re going to hit you and your department with unreasonable requests, ask questions with blindingly-obvious answers, and generally just go out of their way to make your life difficult. It doesn’t help, of course, that even now, IT is treated as somehow ‘separate’ from the rest of the organization in many modern enterprises.
What can you do about that?
According to PC Pro, push for greater integration with the rest of your organization. Don’t take pleasure in operating outside your business, or in being somehow ‘different.’ Particularly with the current trends in enterprise tech (see the above entry), it’s imperative that IT departments start working more closely with their users – and vice-versa.
Hop To The Challenge
There are challenges in every line of work – no matter what it is. In every case, the best thing you can do is make yourself aware of the problems you might face, then take steps to prepare yourself. From there, it’s a simple matter of knowing your business, your users, and – most importantly – your infrastructure.
The rest will sort itself out.