This post was written and contributed by Logan Welliver, Co-Founder of Cloudkick, a Rackspace Cloud Tools partner.
We were very pleased to co-host a webinar with the Rackspace Cloud last Thursday, which gave us the chance to highlight a few examples of how Cloudkick can help users put the focus back on running a successful business, instead of running servers.
Alex Polvi, my counterpart and co-founder, started off with a walk-through of the system, demonstrating how quick and easy it is to go from set-up to monitoring an entire infrastructure, with special emphasis on one of our newest features: Hybrid Cloudkick, which lets users manage and monitor dedicated hardware in the Cloudkick dashboard, right alongside Cloud Servers. Our system makes almost no distinction between the two, giving users the power to leverage api-driven tools against physical machines.
Alex showed how consolidating server management into one place is a powerful asset; coupled with our toolset, it allows users to:
Another focus was how our dynamic, tag-based monitoring system is perfectly suited to the elastic nature of the cloud. In a nutshell, once a user creates a monitor targeted at a specific tag, any time a new server is brought online and has that tag applied, Cloudkick will automatically start monitoring it, building graphs, etc.
Also of note is how a couple views inside the Cloudkick dashboard are reminiscent of open-source tools like Nagios. This is absolutely intentional. We want our users to have all the functionality of tools like Nagios, Cacti, Munin, and others, but without the hassle of installing, configuring and maintaining all these different pieces of software. We’re a drop-in replacement that can be fully configured in a matter of minutes, and requires no ongoing maintenance.
Here’s an abridged transcript of the Q&A session:
Q: I use puppet to configure my machines, are the username and password configurable on the command line instead of interactively?
A: Yes, the Cloudkick agent can easily be incorporated into a configuration management tool like Puppet. Here’s the Cloudkick support wiki page on setting up Puppet to distribute the agent. Additionally, it is possible to use Cloudkick tags as puppet classes, using the Cloudkick External Node Classifier.
Q: Can custom monitors be created?
A: Yes – it’s easy! Custom monitoring works by executing plugins located on the filesystem of the server you wish to monitor. The plugins can be written in any language, the only requirement is that they are executable and output the correct format. To learn more, check out the Cloudkick support wiki page on custom monitoring plugins.
Q: Will you port the agent to BSD operating system?
A: It’s not currently on our road map, because cloud providers don’t use BSD, but if there’s enough demand from our users we will definitely consider it.
Q: When will the Windows version be ready?
A: The Windows port of our agent is in-flight; we have it running as a windows service, there’s an .msi, and it’s currently in private beta. We hope to release it to the public some time this quarter.
Q: Does Cloudkick only work in Firefox, or Internet explorer too?
A: We support the most current versions of all major browsers. Older versions of IE won’t run our more advanced features, like the real-time monitoring visualization, because they don’t support HTML 5 canvas.
Q: Can this be accessed on a mobile device, like a Blackberry of iPhone?
A: Most features are functional on mobile devices. Our main graphing framework is currently built on flash, which obviously presents problems for some devices, but soon we will be transitioning to open technologies that should run on any device.
Q: Is it possible to take corrective actions, for example: provision new cloud servers based on monitored parameter events?
A: Yes, users can specify alerts notifications to fire as webhooks, which means that Cloudkick posts data via HTTP to some URL, so users can write a service that accepts webhooks from Cloudkick, and then take actions based on that incoming data. We send the current status, the reason the alert was fired, and the associated data. We’re also working on integrating alerts more tightly with the Cloudkick agent, possibly allowing it to execute scripts as part of an escalation chain.
Q: Can Cloudkick work with Redhat Enterprise Linux 4?
A: The Cloudkick agent supports redhat 5+; it’s mostly a build issue, rather than a technical one, so if our users express demand, we will certainly address the need.
Q: Most clouds charge by bandwidth; do you have numbers on bandwidth usage for all these real-time monitors? Are there any deals for free or discounted bandwidth for our cloud servers to cloudkick.com?
A: Our monitoring endpoints are located inside the same network as the servers themselves, so it mostly a non-issue. For example, if a machine is on Rackspace, the monitoring endpoint is also on Rackspace, and that endpoint is what goes outside the network to talk to cloudkick.com so we absorb most of the bandwidth costs for monitoring. There are certain edge cases where data travels from a specific cloud server outside the network, but we’ve created a custom, very heavily optimized protocol which minimizes bandwidth usage.
Q: Is the agent open-source?
A: The agent is currently available only as a binary, but Cloudkick hails from a heavily open-source background, and it’s something we care a lot about. Libcloud is a good example of our commitment to this idea. So, we haven’t open-sourced the agent yet, but it’s something we’re definitely thinking about, so it’s safe to say it’s somewhere on the horizon.
Click here to view the entire webinar:
Also, we’ve put together some special discount pricing that’s available to all Rackspace customers until the 15th of May! Take advantage of it here: