This syndicated post originally appeared at Zeus Kerravala's blog.
I saw this recent white paper on the IDG website on why Citrix’s Netscaler ADC beats F5. It’s an interesting read that created a handful of inquiries from end-user buyers, channel partners and investors, but I don’t think it accurately portrays buying criteria. So, after spending hours on the phone, I thought I would summarize the results in this blog.
Citrix did make a few good points on the pay-as-you-grow functionality to scale on demand, but that’s really just an economic argument. The feature parity between hardware and virtual ADCs is a strong differentiator for Citrix Netscaler as well, although, I still believe nothing scales like hardware and the use cases of the software-base ADCs right now are limited to developers and mid market. As they start to work their way into cloud deployments I think that will become a bigger competitive differentiator for Citrix if F5 doesn’t adjust accordingly.
From the deals I have seen though, F5 is the preferred vendor by a wide margin, which explains its 50%+ market share. From conversations I have had with buyers, here are the main reasons why.
- iRules and DevCentral. This is, by far, the biggest differentiator when it comes to F5’s competitive position. For those who aren’t familiar with iRules, it’s a TCL-based scripting language that allows F5 administrators to create custom features. DevCentral is a Wiki-based community forum that enables F5 administrators to interact with one another. Early on, most of the F5 competitors were highly critical of iRules, saying that F5 was offloading feature development onto its customers. While this is true, the F5 community loves iRules. Is TCL the most elegant language? Heck no, but iRules has built this cult-like following that’s hard to unseat. As one CIO told me, once you write one iRule, it’s almost impossible to replace. Additionally, the community is so well structured that many other vendors have used it as a model for their communities.
- Application breadth. The “A” in ADC is for application. The wider the application breadth, the more value it has. F5 has been good and creating application-specific features to help accelerate deployment. The latest example is the support for Google’s SPDY protocol announced at this year’s Interop. F5 was also first with Share Point and Lync support.
- TMOS. This is F5’s operating system. In the early days of TMOS, I think it’s fair to say that the monolithic nature of TMOS made it hard to work with and wasn’t very administrator friendly. The version 10 release of TMOS was a huge leap forward, as it finally modularized the OS. Want to add a security function? Call F5, get a license key and it’s on.
From a performance standpoint, each vendor can claim superiority depending on what kinds of tests are run, but the clustering technology from F5 does allow F5 to scale out very quickly. Overall though, I think both vendors perform well enough that most customers aren’t tapping out all the resources, meaning the techy stuff the vendors like to promote doesn’t matter as much as the stuff I listed above.
Is F5 perfect? Heck no. They are certainly expensive compared to much of the rest of the industry and some customers have claimed procuring products from F5 is difficult (although the company has seemed to have fixed some of the problems). I also think if the world were to all of a sudden adopt software-only ADCs, F5 would be much more challenged by Citrix and Riverbed than by the hardware platforms, but the majority of companies I deal with prefer the hardware solution.
So, again, while Citrix raises some valid points in their white paper, I do think we often overlook why customers buy products from certain vendors. The ADC market is driven by application breadth, which includes the ability to customize the solution. Netscaler is a great product, but sometimes it’s the solutions that customers buy.