Here's a sample of F5 customer comments shared on social networks and forums.
"We've got a handful of large-ish F5s here, and the number is growing. They started as an evaluation project to update our old gear (mostly Cisco), and have stuck around. The networking guys love them, they provide capabilities we didn't have before (GTM rocks!), the app owners can write complex iRules to make them stand on their head, and from a sysadmin's point of view, they just...always work.
I'm generally pretty conservative when it comes to networking gear (and gear in general), but F5 completely wins in my mind."
"Joining the cloud of fans. We've got HA pairs of them all around; we're using them for SSL termination and have used their scripting functionality to solve every single niggling problem we've had with specialized behavior requirements. They stand up to a hell of a beating and are easy to use and administer.
Several thumbs up, would buy again."
"Front-ending your service with an application delivery system like F5's BIG-IP will certainly help with support and troubleshooting. Their DevCentral site is extremely helpful, and the F5 community provides emails assisting with custom code, Podcasts and Youtube videos. I highly recommend them."
"The beauty of F5 is its adhering to Green DC policy i mean to say you must be having different appliances for different jobs but if u look at the feature set of F5 you can find most of the jobs u want ur device to perform can be done by single appliance itself for e.g. Load Balancing with different algorith, DC DR failover, SSL Offload, Out of box functionality, Wan accelaration, SSL VPN, Application security....etc. all this can be achieved in a single box which will save ur invenstment in Rack space, cooling and most important operations expenditure. So if you see overall support list I will definately go with F5 cause I see my ROI getting quickly recovered and TCO is considerably less if combine all features requirement."
"Just finished failover testing a couple of weeks ago between an HA pair of F5s. The failover was so quick that we couldn't actually see a site outage for any of the dozens of sites hosted by the pair. Failover was complete within 3 seconds.
Post-failover analysis indicated that all sites were performing perfectly- and we make heavy use of iRules and WebAccelerator (traffic management scripts and highly customized caching, respectively).
Failing back that same night took, again, less than 3 seconds and again, all sites were performing perfectly.
Love, love, love F5. Their tech support team makes me want to work there."
"I'm increasingly falling into this camp, I think based on current experience I'd choose them wherever I went. I come from a background of using Netscalers and ServerIron LBs, so when we started looking at replacing the home-grown load-balancers at my current job my first inclination was to go with Citrix. I knew them better and had a fair idea how to do what we needed with them (the infrastructure is rather organically grown.) Looking around at our sister-companies and parent company there was a preference for F5 and F5 came back with a competitive quote, just a smidgen higher than for Netscalers, so I rolled that way.
It took a little while to wrap my head around their perception of things, and how to map our infrastructure to it, but once I did it became pretty straight-forward to set up. iRules are just incredible in their flexibility and run fast. It's unbelievable what you can do with them, totally not what I expect from a load-balancer!
The API for the device is also superb and well documented, along with example code. It has allowed me to relatively easily knock up a set of scripts in python to manage the device with, allowing us to split up the irule into components (80+ vhosts, which is just evil in monolithic form), and thus also store them in our VCS!"
"They're awesome pieces of kit, we have seven of them. They take machines out of the pools when the back-ends die, and re-add them automagically when revived. The simple web-based interface allows us to see at a glance the health of the pools, how much traffic we're pushing through it, and allows us to stop traffic flowing to a particular pool member (if we want to just deploy to odd or even hosts).
It's made life so much easier for us that we would never go back to our previous solution (each cluster that wanted load balancing would request their own pound/haproxy front-end).
You should probably consider getting them in pairs, and linking them together for failover."
I would never build a high-use, always-on, production web infrastructure without them.
Loved when I was using them.
They use a TCL based language for the traffic logic, I used the crap out of that feature.
Also, being able to control availability to the backend was amazing. Online code-deploys, zero (external) downtime. Very nice!
I also used the (REST IIRC, twas a perl library) interface to programmaticly add new hosts to the load balancing pools which integrated with my configuration management/deployment system (cobbler post-install triggers). When this was working well I could go from a powered off server, to a new host in the production pool in less than five minutes; with the only manual interaction being the plugging in and powering on of the box.
If it sounds like I'm gushing, it’s because I am. They are more than worth the money if your infrastructure makes enough revenue to justify them.
Disclaimer: The quotes above were taken from independent third party sources, and F5 assumes no responsibility for the content therein.