F5 Friday: A Network Heatwave That’s Good For Operations
The grab bag of awesome that is network-side scripting is, in general, often overlooked. Generally speaking “network gear” isn’t flexible, nor is it adaptable, and it certainly isn’t extensible. But when you put network-side scripting into the mix, suddenly what was inflexible and static becomes extensible and dynamic. In many cases if you’ve ever said “I wish that thing could do X” well, in the case of application delivery it probably can – you just have to learn how.
The how, in the case of F5, is iRules.
iRules is network-side scripting, so it’s executing “in the network” as it were, as data is traversing from client to server and server to client. iRules lets you intercept data (it’s event-driven) and then, well, what do you want to do to it? You can transform it, apply conditional policies, log it, search it, reject it. Being network-side means you have context, and with context you can do a lot of application, location, and client-specific “things” to data and requests and sessions. Being “in the network” means you have access to the full network stack. If you need IP header information, you can get that. If you need application-specific information from within the request or response, you can get that. The entire stack is available to inspect and can ultimately be used to instruct BIG-IP.
iRules figures prominently into F5’s vision of cloud computing as one component of the dynamic control plane necessary to realize the benefits of cloud computing and virtualization. It figures prominently into agile operations and the ability to respond rapidly to datacenter events, and it’s integral to emerging switching and routing architectures based on content and context, such as message-based load balancing and content-based routing.
I don’t often get a chance to cook up iRules myself any more so those who do – and are masters at it – make me just a bit jealous. And it’d be hard to find someone in our kitchen that makes me greener than DevCentral’s own Colin Walker.
He’s done it again, and I can’t say enough good things about this solution. Not only is it hot (pun intended) it’s highly applicable to a variety of uses that go beyond just generating eye-candy for operations. What makes it really interesting is the use of an external service to generate a dynamic view of real-time operations. Colin is leveraging Google’s charting API and relying on the confidence of GeoLocation data based on our integration with Quova to enable a real-time visual display of the locations from which HTTP requests are being received by a BIG-IP. It’s dynamic integration, it’s Infrastructure 2.0, it’s devops, it’s just … cool.