40-Gbps Switching fabrics

chrisone

Geez! now imagine ether-channeling 40Gbps interfaces lol Many vendors including Brocade and HP, sad thing is Cisco had no part in this technology revolution. Fire John Chambers? I guess that direction was away from 40gbps ethernet? I am already looking into HP certification, am i too harsh?

40-Gbps Ethernet Makes Its Debut At Interop -- InformationWeek

Ahriakin

Speed jumps need to be orders of magnitude greater than the previous generation. With the abilities for most vendors to support some sort of L2 bundling, etherchannel etc., attempting to migrate to a speed you can easily reach (and surpass) with bundling is a bit pointless imho. Cisco asked us about 2 yrs back if we had an interest in 40Ge tech and even then (as we were just migrating to 10Ge std.) we said there was no point - we can etherchannel to 80Gbps if needed, the next jump worth trading hardware for is 100Ge. It may be of some use for WAN but I think that's it.
Methinks 40Ge vs. 100ge is WiMax vs. LTE. A quicker solution that will ultimately fail and lead to greater costs in replacement.

CodeBlox

Wow, thats a high throughput. I may be a bit of a noob but wouldn't all those end devices also need NIC's that support that speed? I thought the 10G-baseT was pretty fast. Heh, 40gbs

chrisone

its just not getting any better for cisco.

Cisco Plans Job Cuts, Drops Growth Target | PCWorld

networker050184

Ahriakin wrote: »

Speed jumps need to be orders of magnitude greater than the previous generation. With the abilities for most vendors to support some sort of L2 bundling, etherchannel etc., attempting to migrate to a speed you can easily reach (and surpass) with bundling is a bit pointless imho. Cisco asked us about 2 yrs back if we had an interest in 40Ge tech and even then (as we were just migrating to 10Ge std.) we said there was no point - we can etherchannel to 80Gbps if needed, the next jump worth trading hardware for is 100Ge. It may be of some use for WAN but I think that's it.
Methinks 40Ge vs. 100ge is WiMax vs. LTE. A quicker solution that will ultimately fail and lead to greater costs in replacement.

Yep, I agree with this. Anyone who needs 40G is probably already doing it with a bundle. No point in upgrading hardware to move that to a single interface IMO.

I'd have to say the whole idea behind this from HP and others is just what the OP has came here with. "Oh look Cisco is behind the curve!" I think they are making a smart business decision to push forward to 100G.

chrisone

Yeah 100gb is the frontier, however it doesnt look good when your company is losing profits then you fall behind in a technology race, be it a minor one. HP provides life-time guarantee on its products, i find that amazing! I wish cisco would do the same. With their profits dropping, high priced products, expensive tech support, the need to revamp the company, laying of employees, that is huge stuff my friends. With these recent news bits about cisco you can bet their stocks will continue to plummet.

Dont get me wrong i am a cisco guy but in all honest truth, when i step out of my biased cisco background, it just doesnt look too good for cisco right now.

it_consultant

chrisone wrote: »

Yeah 100gb is the frontier, however it doesnt look good when your company is losing profits then you fall behind in a technology race, be it a minor one. HP provides life-time guarantee on its products, i find that amazing! I wish cisco would do the same. With their profits dropping, high priced products, expensive tech support, the need to revamp the company, laying of employees, that is huge stuff my friends. With these recent news bits about cisco you can bet their stocks will continue to plummet.

Dont get me wrong i am a cisco guy but in all honest truth, when i step out of my biased cisco background, it just doesnt look too good for cisco right now.

I work on a LOT of networks and I cannot find good reasons for people to pay for the cisco brand. Pro curve switches and WG, Sonicwall, some other cheapo firewall is usually the way it goes. Sometimes I can get them to splurge on a lower cost aironet; thats usually only after their dlink fails them too many times. The problem with the competitive firewalls and switches is they aren't really known for failing...and they are cheaper.

billyr

networker050184 wrote: »

Yep, I agree with this. Anyone who needs 40G is probably already doing it with a bundle. No point in upgrading hardware to move that to a single interface IMO.

I'd have to say the whole idea behind this from HP and others is just what the OP has came here with. "Oh look Cisco is behind the curve!" I think they are making a smart business decision to push forward to 100G.

100G doesn't seem that long ago, I was dragging 10base2 around.

tiersten

What Ahriakin and networker050184 said basically. 40Gbps gear doesn't provide enough of a jump in performance to make it worthwhile to upgrade so we're waiting for 100Gbps pretty much. We've got a small amount of 40Gbps gear in specific locations due to some application specific reasons.

it_consultant wrote: »

I work on a LOT of networks and I cannot find good reasons for people to pay for the cisco brand. Pro curve switches and WG, Sonicwall, some other cheapo firewall is usually the way it goes.

There are plenty of good reasons. It depends on whether the brand is important, what you're doing with it, where you work and what they're willing to pay. We generally only use Cisco networking hardware because we're willing to pay for it, we know it'll work properly and if anything goes wrong then we can call up Cisco TAC then start shouting.

We've got Nexus switches in our core and I don't think there are equivalents to that in the HP world. I could be wrong though.

Cisco firewalls are pretty meh however so I do spec out Juniper gear for that as well.

it_consultant

I don't work anywhere that needs the nexus style switches, I don't know if HP has an equivalent. The ISP I work closely with uses Brocade / Foundry hardware as much as they can with a smattering of Cisco. The point I was making is that I have never had a failed ProCurve, if I did they have a lifetime warranty, and they do work correctly straight out of the box. I can't justify the higher price of a Cisco switch. If ProCurves were failing left and right then I could get them to pay for the Cisco name.

If I ran a datacenter where we needed higher quality switching I would lean toward the brocade stuff because thats what the network engineers I work with use and I would want their help.

networker050184

If you are using switches right out the box then I can see why you wouldn't waste the money on Cisco. If you need highly complex features and support then you would think about spending a few extra bucks.

chrisone

not sure if you have used HP switches, I do on a daily basis myself, however they are far from being featureless. They have a lot of huge features that you would find on a cisco switch for half the price. We use them in our access layer and some distribution layer networks. However we are looking into the high end HP core switches as an option for our data center. I dont think it will happen, i still would like a pair of 6500s or nexus.

MickQ

Yep. As much as I like the Cisco stuff, the HP is (for the most part) just as good (lifetime warranty FFS) and is cheaper. Also, the coding for the HP is similar enough to Cisco.

chrisone

yeah the coding is almost identical.

Here is their main option for datacenter switching.

HP A12500 Switch Series - HP Networking

Advanced architecture: midplane, CLOS
6.66 Tb switching capacity, ready for 13.32 Tbps
High-density 10 GbE access with 128 1:1. 512 4:1
40 GbE/100 GbE future access
Redundant switching fabric, power supply, fan tray

Not bad, probably half the price of a Nexus.

All in all , i am only speaking from a switching standpoint, i would still take security ASA and routing ISR routers from cisco over anything else. I am sure other vendors have good routers and firewalls but I trust cisco more in regards of those areas.

Ahriakin

Cisco also have a lifetime warranty - hardware that is. You can take any box you've ever bought legitimately and send it back for repair for free. The extra bells and whistles with Smartnet are for things like TAC, Software upgrades, faster replacement plans etc. Now these are pretty much a must for production gear but for older non-critical systems that basic fallback still comes in handy.

it_consultant

MickQ wrote: »

Yep. As much as I like the Cisco stuff, the HP is (for the most part) just as good (lifetime warranty FFS) and is cheaper. Also, the coding for the HP is similar enough to Cisco.

The major failing of HP networking gear is their complete lack of a certification track that is on par with Cisco. I credit Cisco with being the premier provider of network certifications. Obviously they don't teach Brocade or Pro Curve in the CCNA and CCNP courses. Those people with the interest in networking are going to lean towards Cisco. Most of the network engineers I know got their start in Cisco (because of the certifications) and only started working on other gear when they entered a job that used them. From their they either had respect for or completely converted to a different brand.

chrisone

I see what your saying and i agree about the certs part. The beautiful thing about cisco education is they teach you the standards IEEE non-propietary protocols like OSPF, STP, RSTP, MST, etc. After working several years with cisco equipment you get the idea of how to navigate through command line no matter what IOS you are on. I havent touched a juniper CLI , yet i feel comfortable/confident enough to search my way into most commands they have to configure what i may need.

I have worked with foundry, HP, cisco, Astaro FWs, Sonic wall FWs, and a lot of the commands are very similar in all devices. There is a basic logical way of interpreting commands. They end up having the same logic and categories, sub-categories, well at least from my point of view

I think HP certs have been revamped and they have a new program out now. You can also substitute exams for cisco certs you may already have.