The configuration syntax works as expected, it's flexible but extensible in the right places and mostly has lots of knobs to tune things but without requiring extensive boilerplate to get things to work as expected.
The hardware seems to fit intelligent niches and actually does what it says on the spec sheets.
The system is FreeBSD, while that's a debatable choice in 2024 it's far better than the obscure OSes that others are using for their core systems.
At the end of the day you can always just "start shell user root" and be root on a FreeBSD system. If you're a bit crafty on most Juniper systems you can also run unsigned code if you like. Some Juniper platforms (and not talking about just the huge ones, some of the 1U pizzabox switches) allow you to run a full-blown Linux system alongside the FreeBSD image (because they are actually Linux running FreeBSD on a hypervisor).
JTAC was filled with skilled, intelligent engineers who actually cared and tried to solve your problems. (Not sure today as I haven't had to call them in years). The other TACs I've delt with were focused on call time and ticket handling metrics and would ask obviously pointless questions or repeat information you already provided for a fast close.
I still remember calling JTAC during one outage we had. First L1 engineer after my initial description of the problem basically came back and said something like "that sounds like a really bad outage, please open SSH from 1.2.3.4 and add this public key and I'll login and get the information for L3." Within about 15 minutes I was on the phone with an L3 engineer who correctly diagnosed and proposed a fix for the issue. Amazing support. Not perfect, I have some horror stories too, but, always with people who cared and far less than with C or other vendors TACs.
Very sad day.
Also, sometimes, you just want to throw a tool on there to test something.
Today, I'd love to run stuff like tailscale for access to the control plane remotely.
Mostly, however? I did it because they said I couldn't.
Its not really a bad strategy. If you're going to have a large team of network engineers work continuously on a the Juniper tech stack, they'll get use to it - even if they were raised on Cisco IOS. Juniper stuff works just as well as the rest of them.
It has been years since those features are best served via a general-purpose CPU
Those appliances are nothing but a rebranded servers with some more-or-less interesting software
We bought Juniper gear at the time because nothing Cisco had would work well for us. At least at not any sane price point, and lots of restrictions/gotchas.
Cisco finally got their wind back eventually on their Nexus gear, catching up and run neck-and-neck between Juniper & Arista now.
We continued to buy Cisco for longer than we should have, because "New stuff is surely around the corner", but it's not. Cisco still makes enterprise equipment that can't do IPv6. So we switched to Juniper three or four years ago, and may of our customers are doing or considering the same.
Cisco isn't the dominate play it used to be and is almost never the first choice for new projects anymore.
Not really. Which is why this acquisition price is lower than all their competitors.
> Cisco has stagnated for so long and continue their attempt to push over prices solutions that simply isn't as good or modern as those of their competitors.
Which products, which competitors?
> Cisco isn't the dominate play it used to be and is almost never the first choice for new projects anymore.
Maybe for you, but they are the best in a variety of areas and continue to bring innovation (like Silicon One).
Like any large product company, they have winners and losers, but the existence of losers doesn't assure the absence of winners.