Remember what it was like to drive without a GPS? Sure, it’s possible, but a good GPS takes it to a whole new level. Need gas? A Denny’s Grand Slam? A detour around traffic? You’ve got it! And when the kids start asking “how much longer” you have a precise answer!
Old-school server metrics are like the gauges in your car: They show what’s happening now and can be useful to the driver, but a lot of questions are left un-answered. This is where application performance monitoring comes in: Rather than just checking server stats, APM gives credible, actionable, and user-focused answers about the state of your systems.
System Versus Application Metrics
The metrics collected make up the key differentiator between system performance monitoring and application performance monitoring. Every systems administrator has his own bag of tricks. My own UNIX background leads me to rely on the old-school command line utilities like sar and iostat, but more-advanced applications abound.
Most modern operating systems include a suite of tools for real-time and historical system performance statistics monitoring. Solaris leads the way with the amazing level of detail available from dtrace, but every system offers the basics: CPU load, memory usage, I/O stats, etc. Server monitoring generally focuses on single running operating systems, and the statistics generated usually focus only on a single server or virtual machine.
Data is widely available, but the issue for most systems administrators is what to do with all of it. Does it matter if a server’s CPU is pegged right now? How will that impact application performance and user experience?
Today’s applications are made up of multiple processes on multiple servers. Imagine a web application: It uses a number of web servers, a load balancer, a database, and potentially many other application components spread across dozens or machines or more. System performance metrics can suggest performance problems, but only an application-aware monitor will tell you how user experience is impacted.
The Impact of APM
Once APM is installed, IT starts looking at things differently. They will begin by logging in and checking latency and responsiveness but will be enticed to dig deeper, measuring each step in a transaction or how long it takes to process different kinds of requests.
Large businesses already have sophisticated things like this — they actively measure the performance of mission-critical applications. Quest, HP, CA, and others are raking in profits from APM products today. BMC and Microsoft are also present in this space, as are many newcomers. Each of these vendors is working their own angle, trying to come up with a special useful approach for application managers.
But APM isn’t easy, especially for smaller businesses. These applications cost hundreds of thousands of dollars to purchase, but buying the software is just the start. Actually implementing it takes serious time and development effort, often eclipsing the acquisition cost. And most enterprise-scale APM implementations are highly customized.
This level of expense and effort locks out small and midsize businesses and less mission-critical applications. They just can’t justify the investment! Now infrastructure-focused companies are getting into the market. These offerings are more useful out of the box with less configuration and customization and are targeted at smaller shops.
I recently got a look at just such a product, as SolarWinds presented at Gestalt IT’s Tech Field Day 4. The company is already something of a darling among techies, especially at small-to-medium shops, thanks to their range of free tools and “Thwack” discussion forum. SolarWinds was founded by sysadmins after all, and they take high-end, high-touch, tricky software like network monitoring and bring it to the masses at low cost.
Their new APM product will be released soon, but I got a peak at it during their presentation. It includes pre-configured, canned reports, making it more install-and-go than other APM products. It’s not really competitive with the big boys’ offerings, but brings APM capabilities to the rest of us. Although pricing isn’t available yet, I expect it’ll range from the low-thousand-dollar mark.
SolarWinds offered the Field Day delegates a chance to try out a beta of APM, but I’m not really in a position to give it a fair evaluation. So I asked if I could give my readers the opportunity to try out APM before it’s released. SolarWinds will give you access to the product, support, and a flip camera to record your thoughts and feedback. They’ll presumably use this for marketing and product development purposes. If you’re interested, please mail geek…@solarwinds.com and tell them Stephen Foskett referred you. And please leave me a comment and let me know what you think of the product!
Disclosure: Although this is not a paid or quid pro quo post, SolarWinds sponsored Tech Field Day, which I organize and manage.