Juniper Networks to Combine with HPE | Hacker News Reader

Juniper Networks to Combine with HPE

blogs.juniper.net

147 pointsawat123 comments

Open articleView on HN

Read: "Bill & Dave" by Michael Malone.

Agilent took the heart & soul of HP. Then it split further (Keysight et al)

HP took the printers, which at least used to be a good business.

HPE is the wretched refuse that's left.

> printers, which used to be a good business

Good business as in earning money or as producing customer satisfaction? I don't know about the former, but I do know that HP printers are a horrible user experience these days. The printer is connected via USB, but it cannot be used without an internet connection to their data collection and ink selling cloud nonsense. I will never by anything from that brand again. (I used to be a HP fan since the 1980s.)

Never buy a consumer print device. The enterprise kit from HP is still good.

I print well below 100 pages a year. What would be the price per page using an enterprise grade printer?

If you get a used one, might not be that bad.

But a consumer Brother is probably the way to go. Always search the model number and see what people say.

My $100 brother laser printer I bought 6 months ago has printed everything I asked (dozens and dozens of pages) from Windows, Linux (well supported linux drivers!) and our families mix of apple and android phones.

I have a similar story about mine, but it’s nearing seven years old and just keeps doing what it does.

It doesn’t even have WiFi or AirPrint, but brother has an app that can print to it when you’re on the same network with an iPhone, which is surprisingly convenient.

HL-2170W and iPrint&Scan

Brother is pretty much the only printer brand I still trust. So far, no bullshit of any kind.

Mine died after printing three pages so your mileage may vary.

i got a free unit off of craigslist, nineties HP laser. Ive never changed the toner or added more paper than it came with. I did buy a pi for it to use as a cups server though.

Not sure exactly where the line lies between "consumer" and "enterprise". But, anecdotally, my employer (which is a pretty big enterprise) has small HP laser printers which cost a few hundred dollars (<500) with duplexers and wired network cards. The model is M402 or similar.

They've always worked perfectly on both Linux and Windows and need no maintenance, just remember to turn them on. Most of them are still on their original toner (we don't print that much).

Newer models seem to come with wifi, but I have no idea how well that works.

This is the secret to a great printing experience. Buying a no-frills laser printer, preferably black/white only.

I have the $80 Brother black and white networked laser printer. It’s been great. No hassles. It just works.

At a 100 pages per year, your cheapest option is probably a bus ticket to a Kinkos.

I'll share my example. I bought a Brother DCP-L2540DW (networked, full-duplex B&W laser, scanning with ADF) in 2015 for ~$120 after tax. I've gone through one toner cartridge for ~$50, so total cost excluding electricity is ~$170 for me. I've printed ~3,700 pages so far, so average price per page for the printer has been ~$0.046. I get paper for ~$0.03/page, so total price all in was probably ~0.076/page. This will continue to go down, the printer's current drum life is still >8,000 pages and should last me at least several more years.

With ~100 pages/year that would have been ~900 pages so far. At 900 pages, that probably would have been fine with the included toner, so toner costs don't count. Assuming you turn it off when not printing so energy costs are negligible. That brings it to ~$0.16/page (with my ~$0.03/page paper) printed.

Printing at the FedEx shop is ~$0.23/page for B&W. The public library nearby has B&W printing for $0.15/page. So with ~100 pages/year its cheaper than a chain office shop but more than the library at this point. A few more years will shift that closer and eventually beat the library price.

I've just given up on the economics of photo printing at home. It seems to me you really need to print a lot of photos to compete with the ease of photo printing at most photo stores in the US. You'd need a really good photo printer to match most of the quality of online stores, so you'd need a pretty good printer with nice ink and paper. In-store pickup at a number of places usually means like an hour turn around or so.

Note though that overall my value for my printer is higher than just the printing. As mentioned, it is also a scanner I've scanned ~1,000 items with it.

HP has two tiers of consumer print devices. Bottom of the barrel & "Home, Home Office and SoHo". Second category is good. First one wears down fast.

I almost never have to print and in the rare exception I do I go to the library.

IMO you're best off avoiding products from any company that sells garbage at any tier, whenever possible. They're taking advantage of people and creating large amounts of e-waste.

I have a 16 year old usb printer from them that I bought for 120 euro or so at the time. Still works great. Good value. I buy some toner for it every five years or so and that's about it. It doesn't get used a lot.

Basically, don't buy an inkjet printer (of any type) unless you really need one for photos or graphics work. And then only buy one if that's a regular thing. Otherwise, outsource to professional print shops in your area. It will be better and cheaper.

For normal office stuff, a laser printer is just easier. The toner lasts for ages, doesn't dry out. And if, like me, you rarely need a printer, it's nice when it still works after just sitting there for half a year of not being used at all.

I'll just never buy a printer without an ink tank again.

Ink tanks are better but still not great. If the cartridge had the print head, at least you will replace both if there's a problem anywhere in the cartridge. If it doesn't, such as in all eco tanks, it can clog, especially if not used in a while.

Laser printers are just better. I can't think of a reason for inkjets anymore that's not niche.

Inkjets can print photos

> Good business as in earning money

in the late 80's, when Desktop Publishing was a thing.

I have nothing but good feelings about LaserJet I-6 and the 4000s era circa 1998. I've heard newer models aren't quite as tanks.

Lots of good/reasonable experience with Deskjet 660 and 720C. Although the driver situation for 64-bit windows got real terrible (microsoft had a working driver for server 2003, but insisted on shipping a broken driver for vista/7, so you had to dig up the old one. I think Windows 10 changed the driver model, but I had abanonded the printer for something worse by then. HP support forums could tell you how, but couldn't get it fixed in windows update)

My 4000N from october -98 is still running strong. It only has around 140K impressions so far, so I guess I’ll die before the printer :)

HPE took the PC server business. I'd hardly call that "wretched"?

3par has a less than great reputation

But Nimble has a fine reputation.

I mean, they still offer financing on Itanium CPUs https://buy.hpe.com/us/en/options/processors/intel-itanium-p...

Why does this site try to load Widevine?

Doesn't HPE still have a competent HPC division with Cray and the Chapel language?

I believe all the cray stuff is being spun down and discontinued.

They seem to be pretty represented in TOP500 with Slingshot. Shame if it is dead.

> HPE is the wretched refuse that's left.

Cray tends to disagree...

> Accelerating AI-Native Networking Leadership

I think I just threw up in my mouth.

That was one of the worst things I've ever had the misfortune of reading.

Well someone is diligently using latest version Corporate BS generator which has now included AI* stuff.

You think you did. But if you were using Juniper's AI-Native Networking, you'd know for sure.

I would very much like someone in charge at either HPE or Juniper to explain to me in plain English what that means exactly, in practical terms. I doubt it means anything.

It's as coherent as broadcom's 'new IP'

They are focused on their Mist product line which uses machine learning to troubleshoot and root cause issues.

And it is a fairly interesting product.

Their direction seems to be this framing.. "why would I want to look at a set of graphs/metrics when I can have something tell me exactly what is wrong and make recommendations on how to fix?"

Whether it works in practice or not, who knows. But it is interesting.

> At Juniper, “AI” stands for “All In.”

This is where I vomited.

It's sad, HPE will suck out all the life-force of Juniper; I have had other HP-acquired products and they never got better in subsequent offerings, only worse.

Juniper died today.

We have been testing equipment from other vendors, so that we could replace Juniper in the future, it needed. Those plans just became much more relevant, unless the plan is that Juniper takes the lead on HPE networking.

HPE would have been purchased by Micro Focus is it wasn't a hardware company. It's really sad to see the decline, HP actually made some nice stuff, 15 - 20 years ago.

> Those plans just became much more relevant, unless the plan is that Juniper takes the lead on HPE networking.

Rami (Juniper CEO) is staying and leading.

Famous HP(E) buyouts:

  - Digital Alpha: dead
  - Digital Storageworks: dead
  - Left Hand Networks: dead
  - 3par: dead

And these are only the ones that died in my own datacenter; the full list is much longer.

Aruba seems to have been mostly left alone post acquisition...

Nimble is still shockingly solid

Great let’s just have 2 vendors in every industry in the world with a combined 90% market share.

Why only 90%? For smart phone operating systems it's higher I'd assume (not counting China, that's not a free country anyway). Same for desktop computers.

A major failure of free market economics that is widely ignored by politics.

We don't have a free market economy, we have bought and paid for regulatory pressure and we call it 'free', while the reality is, it's the worst of both free market and planned economics.

Those very market giants have people repeating that it's the regulators, which are their bane, rather than the Republican Party, which do everything to undermine not only all regulation (you can see how people blaming the regulators would serve the Republicans), but every other aspect of government - even tax collection - that would stand in the way of the power of market oligarchs.

The party doesn't matter, government regulation is mostly allowed to exist at the behest of defacto monopolies. It sure as hell isn't the common man making these things happen, nor is it people who represent us in a meaningful way.

It does play into their hands perfectly to blame 'the Republicans' or 'the Democrats', and it does indeed make the problem seem more solvable, easy solution right? Vote out the other guy, problem solved, except when we do that, problem not solved, it just changes the guy who's cashing the checks because the problem is much, much bigger than that (and those that cause the problem are much smarter than that).

I don't know how you can see GOP behavior, in Congress, the courts, and (when they have it) the White House, and overlook the radical anti-regulation, pro-corruption and power policies and actions.

You don't know how, simply because I do not. I just see the manipulation of regulation, which includes, but is certainly not limited to, removing regulation, from all politicians.

From there it's a simple manner of looking up who their donors are, and, it should come at no surprise to anyone, that the money that's behind these people comes from entities that directly benefit from their actions.

> simply because I do not. I just see the manipulation of regulation, which includes, but is certainly not limited to, removing regulation, from all politicians.

You are entitled to your own opinion, of course, but not to your own facts. If you don't know the facts - and if you are interested in regulation, especially those facts - then that's your mistake and will lead to other mistakes.

Ah, no. For all the failings of the series of duopolies, it’s still not even near planned economics. Just think it through.

>Same for desktop computers.

Does it really matter which OEM you buy your Intel or AMD system from? Desktop computers are a market where the 2 top vendors actually have had all the market until Apple entered the scene with its ARM offerings. I'm excited if others will follow with high-end options (Qualcomm and Nvidia).

Oh right. I meant desktop computer operating systems.

But desktop computer chipsets is the same mess. We can be glad already that there is competition between 2 vendors (1). Some years ago it was probably over 90% Intel.

Servers is not fundamentally better, just with slightly different market shares.

And everything is x86, which is certainly not the optimal solution from several aspects. I wonder whether that would be still alive with some healthy competition between at least a dozen of vendors.

Edit: (1) 2 plus Apple, which is better than only 2, but sti not fully free competition. Few people choose Apple because of their chipset.

Except the big cloud companies have been replacing more and more high end gear from vendors like juniper with in house designs, which are gradually percolating out. High end routing and switching gear is gradually becoming commodity.

> a combined 90% market share.

I was under the impression Juniper market share had been falling consistently. But it sounds like there is a market where they are still doing well, what is it? Enterprise something?

Juniper is still decently popular in their core routing business. On k12 and small enterprise they've had good fortune with their Mist acquisition. I think the latter was honestly the main driver for the merger.

I think GP was trying to say a few network companies are buying everything up not that Juniper had a majority share itself though. Look at the merger graphs for Cisco, HPE (adding in Juniper's now), and Extreme and a lot of the diversity that used to be "not Cisco" and became alternative growth is going back to being a few companies with a stranglehold on share again.

Ah, that interpretation makes more sense than mine, thank you!

I work in a primarily Juniper environment, and I'm not that familiar with HPE (Did they previously have anything in the networking space?). It's sounding pretty doom-and-gloom in here, what should my level of concern be?

I would be pretty bummed to change vendor. Junos is probably my favorite networking OS. The CLI is comfortable, BSD is never too far away, and the config structure is gorgeous. IOS (and similar) can really be a pain to read by comparison.

HP had their own ProCurve networking and they also bought Aruba.

We don't know what HPE is going to do to Juniper specifically, but in general many acquisitions either raise prices or simplify the product line by canceling some products. In this case HPE may cancel some Aruba products and replace them with Mist equivalents.

They also inherited 3Com from HP and the range of H3C switches were far superior to ProCurve for many years. I think they only offloaded their stake last year.

I work in HPC, and we saw the quality of support completely plummet after HPE bought Cray.

I work at a network VAR partnered with both of these companies, but these thoughts are my own assumptions. HPE didn't immediately implode Aruba, I don't think they'll immediately implode Juniper either. That said, I don't think it's a good move as a fan of Juniper myself. I think Mist/AI led the desire for the acquisition, particularly with their performance in the lower end market last year vs HPE Aruba type products which usually win for higher end enterprise. I'm cautious about how they'll handle the other product lines, particularly on how much the care about them. Obviously nothing is disappearing overnight though, this will be something that shakes out over years.

Juniper has no cohesion in their product offering, so fits well with HPE who also fail in this regard. They had all the pieces, ex/qfx/srx/128t/mist/aapstra/mx/ptx to build any network you needed, but no ability to fully integrate all of the things. Surprising that HPE acquires them considering they already competitors across half of those things procurve/silverpeak/aruba/plexxi.

Happy Juniper customer since 2010. Across SRX, EX, QFX lines. What to do now, what to do...

Arista or whitebox switches, depending on how big you are and your appetite to DIY your network as code vs DIY writing configuration.

What is a whitebox switch?

A generic switch motherboard, in a generic case, with a generic PSU.

There's a boot standard, so you can install a range of different Linux-based OS distributions.

Similar in spirit to a whitebox PC, hence the naming.

Google has a zillion links.

https://en.wikipedia.org/wiki/White_box_(computer_hardware)

https://en.wikipedia.org/wiki/White-label_product

It's decoupling the hardware from the software of the switch.

Historically, you buy a Cisco switch, you run IOS (or NX-OS).

You buy a Juniper switch, you run JunOS.

A company called Cumulus came around (https://en.wikipedia.org/wiki/Cumulus_Networks) and changed a lot of that (they aren't solely responsible and there are other reasons why but they are very important here).

At the same time, instead of each switch vendor making their own chips, Broadcom started making generic chips which they would sell to anyone (Lookup Broadcom Trident), so, many switch vendors started using those instead of custom silicon.

At the same time, many people started talking about network as code and doing overlay networking things with control planes that didn't fit into the normal models for how these switches worked and were configured (configuration and routing changes which happen in milliseconds over custom protocols instead of ssh to switch and "configure, set route ..." etc etc).

The convergence of this was that the network hardware and network software got decoupled and commoditized. You could buy a Juniper OCX (the OCX line was specifically for this) and install Cisco NX-OS on it if you wanted.

Broadly this was a really good thing for the networking industry IMO.

Also, all of the above generally only applies to the what people refer to as switches (though they can and do route), and not routers (think BGP edge routers) proper (which can and do switch) and are broadly still custom silicon ASIC with much more bespoke and advanced feature sets and can only run the vendor's software.

Whitebox switches for the win. Last datacenter I worked in we ripped out all our juniper, cisco, and mellanox gear. Went all in on cumulus. The fact that everything is linux made automating everything a breeze. Templated switch configs, automated deployments and upgrades, virtualized switch OS to test config changes via CI.

> What to do now, what to do...

csco beckons thee

> Across SRX

Good lord. I don’t know whether I respect or pity you.

There used to be a saying at Juniper: If there was an issue with MX, it was likely a config issue. If there was an issue with SRX it was likely a bug.

Was incredibly accurate more often than not.

I miss Netscreen.

I have an SRX1400 cluster with nearly 11 years of uptime. But this is the bare metal BSD-based firewall. When later, Juniper started adding a hypervisor layer on SRX platforms, and I started having problems.

Another one bites the dust

Im getting nineties Nortel Bay Networks all stock flashbacks.

The Company that Broke Canada - BobbyBroccoli https://www.youtube.com/watch?v=I6xwMIUPHss

They already acquired Aruba in 2015 (https://investors.hpe.com/financial/acquisitions)...

I was always a fan of HP and 3com for networking over Cisco. It was easy to keep gear assigned to specific tasks. This is our core switch. This is the core router. Pricing kept this amenable to management. Juniper was never "in the budget" as a viable alternative. I'm currently at a Cisco shop that purchased their first Juniper switch 12 months ago for Equinix cross-connects to the cloud. Should be interesting to see how this plays out.

HPE != HP

The HP you’re thinking of is now Agilent.

And this isn’t going to go well. Look at the downward slope of Aruba since HPE acquired them.

and all the storage vendors they have aquired over the years. LeftHand, Nimble, 3par....

From the press release[0]:

"The combination is expected to achieve operating efficiencies and run-rate annual cost synergies of $450 million within 36 months post close."

Not that such layoffs are unexpected.

0: https://newsroom.juniper.net/news/news-details/2024/HPE-to-A...

HPE is in the same boat as Juniper in terms of market and mindshare. I guess they want to see what it looks like to merge I guess

I remember working with Juniper equipment a decade ago (I didn't like it because I've traditionally worked exclusively with Cisco equipment), but I haven't heard anyone in my circles mention Juniper since. Are they considered a niche vendor today (similar to Novell in the 2000s)?

I remember Juniper making news way back when by using FreeBSD as the OS for the router and that was supposed to have saved them a bunch of development resources and earned a bunch of nerd cred.

My work has been buying Juniper switches lately since CISCO has largely priced us out.

We switched to Juniper during COVID because the delivery times were like 9 months for Cisco

Having worked with every vendor under the sun for networking equipment across my career, I enjoy working with Juniper by far the most.

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.

I'm just curious, why would you want to run unsigned code on a Juniper device? What is the use case?

I was doing a number of unholy things at the time, effectively monkey-patching and syscall interception of some of the daemons for "seemed like a good idea at the time" reasons.

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.

Custom collection daemon or other type of agent.

Considering how annoying SNMP is to use and how terrible the alternatives are, that would have been a very good use case.

Companies go with Juniper because they can outfit a data center at half the price of Cisco. NYSE did that a decade or more ago with their Mahwah data center.

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.

But Arista is probably half the price of Juniper so that doesn't make much sense.

But Arista only has switching hardware and a few other specialty devices (L1 switches from their Metamako purchase). Vendors like Juniper offer load balancers, firewalls and everything else you need for managing a data center or company.

Load balancer, firewalls and everything else you need for managing a data center or compagny are virtual machines

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

True, but a lot of companies want to source it all from a single vendor so they can talk to one entity when dealing with configuration/integration issues. Its not how I roll but it is how a lot of big companies operate.

The first time i did a commit, knocked myself off the system, and it automatically rolled back its config 5 min later because I didn't do a 'commit confirm' was kind of mindblowing.

Ha! I laugh because I've done the same exact thing... It is a paradigm shift coming from Cisco to JuniperOS.

For a very long time, Juniper lead over Cisco for performance and features. After the run of the original Catalyst switches, Cisco was floundering, resting on their past successes without really pushing anything.

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.

Agreed, WARP mode on the Nexus does give some pretty nice port-to-port latencies (for a L3 switch - L1 gear blows them all away).

Juniper is pretty big, and probably the company most of the Cisco users are turning to, in my experience. 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.

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.

> Juniper is pretty big, and probably the company most of the Cisco users are turning to, in my experience.

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.

I've worked with a large number of vendors and I prefer juniper. They're very easy to write software against.

Ahhhh this explains why they (Juniper) have been a nightmare to deal with lately.

Any good leads on some “cheap” Juniper 10Gbps switches?

Ebay

and now juniper will perish !!

the enshittifcation and conglomeration of all things continues.

The reverse happens too, HP split into HP and Agilent.

That happened in the 90s

June 2, 2000

Soon it will all be Taco Bell.

Don't you mean Ma Bell, in this context? ;)

Nope, all Bells will be Taco Bells after the Franchise Wars. I wouldn't trust any other restaurant for my telecommunications and networking needs.

Case in point, their 7-layer Burrito [1] offering maps perfectly to the OSI Model [2]. It is fully self-encapsulating (though occasionally leaky), and operates bidirectionally with full duplex support for correctly-configured clients.

[1] https://tacobell.fandom.com/wiki/7-Layer_Burrito [2] https://en.wikipedia.org/wiki/OSI_model

This has the potential to be good. HPE has mostly let the (superior) Aruba line grow as they were, and TBH the CX stuff is quite good. They don't have any enterprise class routing/firewalls. Their AP hardware is great but Central can be frustrating. If they can put the pieces together without fucking things up, there's a good product line up here at a time when people are sick and tired of Cisco's licensing antics. TBD...

HPE took over Aruba wireless and Nimble storage. We used both products. They didn't fuck with Nimble support much. As of a few years ago, it was still great.

Aruba wireless...oh man. The support was so terrible.

We had some high availability controllers go out of sync. Couple hours on the phone, no resolution. Had a bunch of back and forth, another couple calls, nothing. It was sort of funny (and sad) watching a different 'tech' try the same scripted steps, over and over again.

Finally I was able to get the issue escalated, and the next guy I talked to seemed to really know his shit. Made some fundamental changes that should have never been that way in the first place, he got one controller back on, and then in the middle the next controller, he fucking bailed off the call and left us in the hands of a new guy.

A new guy who couldn't fix it. He actually broke it in a new away, and then our maintenance window closed. Two more maintenance windows later, nothing fixed, same useless drones trying the same crap, I just gave up.

I ended up fixing the last bad controller myself. My organization was paying almost my salary per year for this 'support' contract. What a joke.

Reply on news.ycombinator.com