Amazon Web Services to Enter Networking Equipment Market

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
22,060
Cisco shares were down 5% and lost $11 billion in market capitalization after a report circulated that Amazon Web Services is primed to enter the lucrative network equipment market. It is rumored that Amazon will sell their white-box switches for 70% - 80% less than comparable Cisco solutions. These white-box devices will feature built-in connections to Amazon Web Services. The software running on these devices is open-source and the hardware is unbranded to lower the cost. Another upside is that open-source software allows for more customization than proprietary vendor software.

The networking devices will consist of open-source software and unbranded hardware known as "white-box" switches and come with built-in connections to AWS cloud services, such as servers and storage, the Information said. Amazon Web Services could price its white-box switches 70-80 percent less than comparable switches from networking giant Cisco, the report said, citing one of the people with direct knowledge of the unit’s plan.
 
Wonder if this will be cheap enough for Prosumer. Im using a bunch of Ubiquiti stuff currently at home and like it but would like more options.
 
40/100GB switches for pennies on the dollar? Will take 10 please.

Pennies on the dollar is still new car pricing for that level, though...

Wonder if this will be cheap enough for Prosumer. Im using a bunch of Ubiquiti stuff currently at home and like it but would like more options.

Me too- though I mostly just wish Ubiquiti would get their Unifi stuff in order and up to snuff with their other lines (USG and L3 switching, I'm looking at you).

For now, I've been tinkering with used crap off of Ebay in addition to my UBNT stack...
 
Uh oh Cisco's in trouble. If amazon sells network devices with an open source os on it, and doesn't charge a crap ton of money in order to use their products.

Cisco's got some great hardware, but when they charge you for "features" and "support". You buy a cisco router, oh but you want to use advance features such as routing bgp? thats gonna cost you. I've always hated cisco's ios rip-off feature stack and support charges. That's why cisco is losing market share.
 
I just want decent managed lower cost 10gbe switches that I can use for my home lab which isn't crap from ebay.

This.

Why are 10GBase-t switches and cards still so expensive?

Oh hell no. I'm not going to let Amazon monitor and monetize information about my use of the Internet.

And this.
I don't want any "cloud" connected devices running my network.
 
I just want decent managed lower cost 10gbe switches that I can use for my home lab which isn't crap from ebay.

What's lower-cost?

I just ebayed an Aruba (HPE) 2500S that has 4x10Gbe SFP+ ports to dick around with, but I'm looking more closely at Ubiquiti's 16port 10Gbe switch that comes in Unifi and Edgeswitch flavors starting at ~US$550...
 
Wonder if this will be cheap enough for Prosumer. Im using a bunch of Ubiquiti stuff currently at home and like it but would like more options.
That is where the Unifi XG series comes in, it is also 50-70% of comparible Cisco solutions.
 
Oh hell no. I'm not going to let Amazon monitor and monetize information about my use of the Internet.
If it is open-source, then it may allow community-built modified variations of the stock OS. Pick and choose what features you want.
 
What's lower-cost?

I just ebayed an Aruba (HPE) 2500S that has 4x10Gbe SFP+ ports to dick around with, but I'm looking more closely at Ubiquiti's 16port 10Gbe switch that comes in Unifi and Edgeswitch flavors starting at ~US$550...
Fiber is cheaper and more consistent than 10GBE, I am not sure why people keep beating on this 10GBE horse.
 
If it is open-source, then it may allow community-built modified variations of the stock OS. Pick and choose what features you want.
"Open source" just means the source is available. It doesn't mean you get to modify the installed code or control its features.
 
As someone who sells networking equipment that competes with Cisco, I don't think they are shaking in their boots. People buy Cisco because they are infatuated with the brand or they are the safe choice that won't be questioned if something goes wrong. There's been cheaper alternatives that are just as good for a long time now.

Even if Amazon legitimately competes, which I have my doubts, Cisco would still likely have several years to develop a competing solution.

Winning over hobbyists and home labs, maybe. But this is just a late to market SDN solution that won't take off with enterprises, reguardless of price.

I get the stock market shifts on the Whole Foods and pharmacy moves. It's perfectly logical to believe Cisco is trending down, but sure as shit not because any threat from Amazon.
 
Sorry, I'm mixing it up; I do mean fiber or DACs at 10Gbit, not RJ-45 and Cat6/7 :).
Oh in that case look at using 10GB SFP+stacking cables, if they are being used short distances you can get them for as little as $20. Much cheaper that way, longer distances then C2G does pre done fiber that is price comparable to copper and Mikrotik and a few others do 850nm multimode transevers for $50 a pop.

Also keep an eye out come September for second hand networking equipment. Government entities (in Canada at least) auction them off during upgrade cycles. I’m going to be posting a bunch of Dell L3 gigabit PoE switches, all of which have 2 or 4 10GB SFP+ ports. And I know a number of other sites that will be putting up a lot of Pro Curves as well as they fall out of their official support cycles.
 
As someone who sells networking equipment that competes with Cisco, I don't think they are shaking in their boots. People buy Cisco because they are infatuated with the brand or they are the safe choice that won't be questioned if something goes wrong. There's been cheaper alternatives that are just as good for a long time now.

Even if Amazon legitimately competes, which I have my doubts, Cisco would still likely have several years to develop a competing solution.

Winning over hobbyists and home labs, maybe. But this is just a late to market SDN solution that won't take off with enterprises, reguardless of price.

I get the stock market shifts on the Whole Foods and pharmacy moves. It's perfectly logical to believe Cisco is trending down, but sure as shit not because any threat from Amazon.

Yeah, that's exactly what stopped other startups who've been "competing" with Cisco for years.

Thing is, this isn't some obscure, dinky startup. This is Amazon. They have a heck of a reputation already, with alot of marketing power. Alot of businesses use AWS too, which buys them a little trust right off the bat.

I think you're partially right, but I also think Amazon will win over more customers than other competitors have.
 
As someone who sells networking equipment that competes with Cisco, I don't think they are shaking in their boots. People buy Cisco because they are infatuated with the brand or they are the safe choice that won't be questioned if something goes wrong. There's been cheaper alternatives that are just as good for a long time now.

Even if Amazon legitimately competes, which I have my doubts, Cisco would still likely have several years to develop a competing solution.

Winning over hobbyists and home labs, maybe. But this is just a late to market SDN solution that won't take off with enterprises, reguardless of price.

I get the stock market shifts on the Whole Foods and pharmacy moves. It's perfectly logical to believe Cisco is trending down, but sure as shit not because any threat from Amazon.


Ya pretty much. They also offer a very large solution set. Most other companies don't offer everything but with Cisco you can get your routers, switches, DWDM units, CMTS, etc, etc. Plus there's vendor support. Many vendors don't wanna fuck around with lots of different switches to support on their products and Cisco is usually the company they choose. Like if you get a NetApp cluster and you want NEtApp's support, you need to have one of a few specific Cisco switches as the cluster interconnect network. They aren't going to go and support whatever shit you hook up, because they don't wanna troubleshoot network issues on the cluster side on something they don't know about.

But ya if price were the only thing, well then Juniper or another company would have killed Cisco long ago. Right now you can get Juniper gear that performs as well as Cisco gear for less money, and they've been in the market a LONG time so a well established player. Yet for all that, they are a fraction of Cisco's market share (Cisco is like half the switch market, Juniper is like 10% or less).

Then when Broadcom started making really high end switching chips, all kinds of companies cropped up selling switching gear. Dell, Aruba, Arista, etc are all switches that are Broadcom chips on Broadcom designed boards. For that matter if you want "white box" switches companies like FS.com will sell you Broadcom based switches for real cheap, of questionable quality.

We'll see, but I don't think they'll actually make a lot of heardoom. They'll probably chip away at Cisco's market some, but I doubt a whole lot.
 
And you guys thought Google was going to take over the world.
 
Google has long since exited the aggressive cornering the market phase and entered protection mode. So maybe quieter, but no less dominant. They just want to protect the status quo.

Amazon took over the distribution market, but there's very little profit there, it does however position them well for the aggressive moves they've been taking. With AWS footing the bill. And of course their execution has been phenomenal.

Switching doesn't really figure into that, another reason I don't think it's going to be a big play.
 
Amazon's cloud services took a lot of people by surprise. Amazon's cloud isn't radically different from a remote server farm, but Amazon did a clever bit of marketing magic to make time sharing at the enterprise level look like good business. More importantly, Amazon didn't screw it up - they had the hardware, software and support in place to prevent a stumble as the business grew.

Now that the cloud is here and growing people are beginning to see how much more it could do. Cloud services are going to grow, especially when companies contract with Amazon to rebrand Amazon's cloud under their own product name (such as game streaming, where the product is the game and the customer has no idea it is running from Amazon's hardware.) We don't have the infrastructure for the next big leap in the cloud, but this announcement seems more about Amazon creating a complete, Amazon-optimized and Amazon-centric package for their internet services than Amazon looking to compete with Cisco.
 
Should be a wonderful deal for those companies that have already sold their soul to AWS. The built in spy-o'matic features won't be any worse then the ones already in AWS. For the rest of us, just make sure your non-Amazon edge device has the proper block rules for the white box equipment.
 
An acl on an interface is really no substitute for deep-packet inspection, so this and a real firewall. Help to pre-filter and apply the Bogon list, but I'd still have a firewall. I could see maybe if you were connecteing to AWS, but still. And support? Lot's and lot's of cisco support/experience experience out there, not so much for this kind of open source installation/support.
 
Uh oh Cisco's in trouble. If amazon sells network devices with an open source os on it, and doesn't charge a crap ton of money in order to use their products.

Cisco's got some great hardware, but when they charge you for "features" and "support". You buy a cisco router, oh but you want to use advance features such as routing bgp? thats gonna cost you. I've always hated cisco's ios rip-off feature stack and support charges. That's why cisco is losing market share.

Cisco has been in trouble, which is why they have moved out of hardware. You can build or buy what Amazon is proposing currently with cisco or juniper overlayed. Now, it’s a disparate environment, Amazon can do this on a massive scale. I mean, we have vSwitches etc, you can do what you want. What worries me is this; where are they being built? Who is building them? What firmware is it? To get this low of cost, my best guess is China...and those in the know or who follow these things know what the Chinese can do when they are buildings the ICs and ASICs that run this stuff.

I will boycott out of principal. Amazon is too big, and only growing. Shipping, pharmacies, our food, home goods, where will it end? Trying to take over the entire networking infrastructure does not sound like a good plan for us as citizens.
 
I like these but I think they are essentially baked AWS portals for easy infrastructure. Outside of that I don’t think we will see these be in home labs.
 
The commodification of every scrap of our lives, wow.
Well no, it’s just for those who rely strictly AWS. If your business is relying on AWS you don’t exactly need CISCO software running hardware to be effective.

There are still tons of other hardware manufacturers out there.
 
Fiber is cheaper and more consistent than 10GBE, I am not sure why people keep beating on this 10GBE horse.

Fiber is great if you have a large computer room or multiple wiring closets to connect.

10GBase-t is a better solution for smaller businesses, since it uses inexpensive patch cables and is backwards compatible with GB ethernet.

If my main 10GBase-t switch dies, the ability to simply move the cables to a spare 1gb ethernet switch is important.
As for "why not have a spare 10GBase-t switch", it's not in the budget since even a single switch & server cards is currently too much.

With the backwards compatibility, I can buy a switch and cards for the heaviest used servers, and gradually upgrade the others as they are replaced or as needed.
 
Fiber is great if you have a large computer room or multiple wiring closets to connect.

10GBase-t is a better solution for smaller businesses, since it uses inexpensive patch cables and is backwards compatible with GB ethernet.

If my main 10GBase-t switch dies, the ability to simply move the cables to a spare 1gb ethernet switch is important.
As for "why not have a spare 10GBase-t switch", it's not in the budget since even a single switch & server cards is currently too much.

With the backwards compatibility, I can buy a switch and cards for the heaviest used servers, and gradually upgrade the others as they are replaced or as needed.

See all the buildings I work in are all wired in 5E so 10GB isn’t going to work on that to begin with, I don’t think I have ever worked in a site that was wired with 6a. The only equipment I have ever seen that came with a 10GbE was some Buffalo NAS units. Either way it was far cheaper for me to call up Leviton with the lengths I needed and have them prep me the OM3 than it was to work with copper. So all servers and switch interlinkes are fiber.
 
See all the buildings I work in are all wired in 5E so 10GB isn’t going to work on that to begin with, I don’t think I have ever worked in a site that was wired with 6a. The only equipment I have ever seen that came with a 10GbE was some Buffalo NAS units. Either way it was far cheaper for me to call up Leviton with the lengths I needed and have them prep me the OM3 than it was to work with copper. So all servers and switch interlinkes are fiber.

Don't need 10GB to the desktop (at least for now), just between the servers and switches.
I have 3 racks at one office and a single full rack at the other.
I can use a single 24 port 10GBase-t switch as the backbone, connecting the servers and the GB switches that connected to the desktops & other equipment.
(will also be replacing those switches with ones that have a 10GB uplink since these switched are almost 10 years old).

No need to mess with fiber when I can simply run cat 6a or cat 7 patch cables.
Also allows for gradual upgrading of the servers, since 10GBase-t is backwards compatible with the existing gigabit Ethernet.
 
Back
Top