Newsletter:2/The Server is Down but Google Where is the Server

From Zenoss Wiki
Jump to: navigation, search

The Server is Down but Google Where is the Server

The Google Maps Portlet (GMP) has been a part of Zenoss Core for over six years, and amazingly no other open-source monitoring tool offers anything like it.
Article2b.jpg

How Zenoss Core’s Google Maps Portlet Keeps You From Feeling Blue

The Google Maps Portlet (GMP) has been a part of Zenoss Core for over six years, and amazingly no other open-source monitoring tool offers anything like it.

In fact my buddy “Tupelo” (not his real name) says he couldn’t live without Zenoss Core or GMP. A former system administrator, Tupelo owns and runs a velvet painting reproduction house called “Velvis.”

Velvis uses four small data centers: one in the Memphis headquarters, one in Hollywood, one in downtown Los Angeles, and one in a colocation facility in Reyjavik, Iceland for backup and redundancy. Let’s look at how GMP has simplified Tupelo’s IT life and protected his business.

Finding the Catalyst (Switch)

In late July, the system administrator running the Memphis data center added a Cisco Catalyst 2960 Series switch to handle the anticipated increase of traffic. Along with Christmas, August has always been Velvis’ biggest month because it’s the anniversary of Elvis’ death, and over the last few years customers increasingly have bypassed Velvis’ partners to buy directly from the Velvis website.

Zenoss Core immediately modeled the device as being part of IP address “192.168.65.72” but memorizing a series of numbers (let alone four of them) is a hassle at best. Luckily GMP lets you assign a location to each IP address. Even better, you can do so using a street address, rather than having to figure out the GPS coordinates.

Velvis IP address “192.168.65.72” corresponds to the street address of the Memphis data center, showing up on as a green point on the Zenoss Core map. Thanks to this feature, the Memphis system admin was able to discover and label the Cisco switch, drag it from his list of devices in his Infrastructure tab, and drop it on top of that green circle representing the Memphis data center. When he wants to see a list of devices associated with that data center, he can easily bring them up within his Zenoss Core dashboard.

Maintenance in Reykjavik

Because Velvis relies on the Reyjavik facility for failover protection, Tupelo’s admin team needs to set up frequent maintenance windows to monitor the site. However, Velvis’ Zenoss Core installation is in his Memphis data center, and Core only follows the time zone of that primary installation. How does Tupelo schedule that two-hour maintenance window to run every Sunday at 1 AM GMT, when his installation officially is on CDT?

Fortunately GMP provides a visual representation of time zones, and Tupelo’s IT team easily determined Sunday at 1 AM GMT corresponded to Saturday at 7 PM CDT. At that point they realized this scheduled maintenance window would conflict with increased U.S. traffic generated by drunk Elvis fans on a Saturday night. With Tupelo’s blessing, they set up that maintenance window for 9 AM GMT/3 AM CDT, when most of those customers are sleeping it off.

Avoiding a Velvet Crush

Elvis’ Death Day (August 16) is Velvis’ busiest shopping day of the year, and this year it happened to fall on a Friday. So imagine Tupelo’s distress when he saw a red dot over Los Angeles on his Zenoss Core dashboard map that Thursday evening.

Because GMP automatically correlates events and statuses to locations, Tupelo could see the outage at a glance. Tupelo clicked on the Zenoss Core Infrastructure tab, and under Location, accessed a second map to drill down to his sublocations.

The Hollywood data center was the culprit. Tupelo found the device causing the outage and interrupted his model/actor/sysadmin, or MAS (yes, this hyphenate does exist in Hollyweird!), in the middle of the latter’s one-man show “Do VMs Dream of Physical Enclosures?”

MAS left his audience of two to fix the problem before the crush (pun intended) of buyers overwhelmed Velvis’ customer-facing websites. And by that Sunday morning, coincidentally around the time maintenance in Reyjavik was about to commence, Velvis was sold out of every black velvet Elvis painting it carries -- even the ones depicting “fat” Elvis on the crapper.

Turning Red to Green

The Hollywood MAS quit the following week, citing commitment to his art, and Tupelo has since hired a responsible replacement.

But Tupelo still shudders at what would have happened if Velvis hadn’t been using Zenoss Core and GMP to monitor his infrastructure. Velvis’ IT team would have learned about the outage from an angry customer and probably spent even more time isolating the device that caused the problem.

“Without that red dot, I stood to lose a lot of green, if you know what I mean,” Tupelo told me. I do, Tupelo, and I’m guessing that ability to see that red dot and turn it green has kept many a Core user from feeling blue.