the instant ?
really? Maybe if you are in an ultra high security industry, otherwise generally speaking if it ain't broke no need to fix it. I actively held onto one version of code on my switches until I could no longer get support on it because it eventually became end of life (having performed no major updates on the switches in the previous 3 years).
My current switch code has a couple of known issues with it that are fixed in newer versions - but for now (and for the past year) I'd rather live with those known issues then think what new issues might be introduced by the new code. The features I use on my switches really haven't changed in 11 years now.
No I don't use Cisco, and won't touch Supermicro.
Same goes for other things too, the ONLY reason I upgraded to vsphere 5.5 from 4.1 was because 4.1 was past end of support(by about 6 months). Fortunately I don't have to open vmware cases very often. I have no intention of upgrading to vSphere 6 before 2017. 5.5 does everything I need just fine. I had some early issues with some VMs hanging and having to be killed on the ESXi host themselves (first time I'd ever seen that in vmware myself), but that problem seems to have resolved itself at some point.
My Splunk version has been end of support for probably 2 years(yet we still pay for support). I was burned BADLY by splunk 3->4 upgrade years ago and now am paranoid about future upgrades, the last major splunk issue took more than 1 YEAR to resolve(not related to 3-> upgrade that was probably 2 years of critical issues), and as things stand now it WORKS good and nobody has complained in at least 2 years(since the last issue was resolved). No outstanding issues, no pressure to upgrade, it's on the roadmap of things to do just keeps getting pushed because there are more important things to worry about.
My production load balancer software went end of support in October, there is a newer version and obviously we are under support - the newer version has a critical bug in layer 7 mysql load balancing that is preventing us from upgrading, been working with support on it for 5 months now they still don't know the cause(prior to that a stupid licensing decision the vendor made in versions between what I have and the absolute latest prevented upgrading, the licensing decision was reversed in the latest code). My team used to hate the Java UI in the current load balancer version that we have, then they saw the new all HTML5 UI and they hate that even more they want the Java stuff back (well technically they haven't lost it yet in our production load balancers just the non production ones).
Oh and my latest shiny DL380Gen9 boxes have a critical issue where they can take upwards of 2 hours to boot(and at least 2-3 attempts to get them to POST entirely), my earliest systems had to have their system boards replaced because they kept crashing on POST. HP eventually figured out the cause but does not have a solution for me (their solution is to disable boot from SAN but I boot from SAN so that is not an option). At least they fixed the firmware so the system boards don't become unusable.
http://h20564.www2.hpe.com/hpsc/doc/public/display?docId=emr_na-c04822613&sp4ts.oid=7500984
fortunately I don't have to reboot often, and these systems I don't plan to reboot for at least 6 months.
I used to run a full layer 3 switch at home for a few years 48-port 10/100, though it was obviously very loud(even after I replaced the fans) and drew a lot of power. I used 2 or 3 VLANs on it, ditched it maybe 8 years ago for small netgear switches(the little metal ones). No VLANs, nothing fancy. I do have a separate network zone for less trusted things which is wired into my Soekris OpenBSD firewall, so those things like my PS4 are connected to that zone(along with wifi), and have no access to my internal network.
The more time goes on the more I just want shit to be stable, there are very few new features that I care about anymore. Maybe I'm just getting old.