linux router vs cisco router

Hello!

i'm new to the networking world, i have 6 month experience with cisco routers and switches and recently i got my CCNA.I am wondering...what is best? linux router or cisco router; the major drawback to cisco products is their price...and on the other hand linux is free, you only need a server ( which is very cheap) if u use it only for routing and switching.

What do you think guys?

I'm sure that there are a lot of experienced network admins here, who can perhaps share their knowledge...with us the little guys , who are at the beginning of their carrer!

thank you in advance!

Hello Gwiz,

As a linux fan I've found my self many times wondering why would anyone prefer to use an expensive piece of embedded equipment, when he could do so much more at little to no cost, using free, unix-like operating systems. However I have been pointed out that there are cases where embedded hardware is a better option: in mission critical environments and in environments where it is not possible for the personnel to handle the extra administration effort needed by such solutions.

Both choices have some advantages and disadvantages, what is better depends on what priorities you set on the following factors:

- Reliability
- Features
- Knowledge requirements
- Cost

Reliability refers to both software and hardware behaving as expected for extended periods of time. Generally, embedded hardware like CISCO tends to be more reliable than PC hardware. The fact alone that most PC-based routers would depend on a hard disk drive compromises their reliability, since mechanical parts have much shorter lifespan than electronic parts. Furthermore, unless each part of the PC was selected carefully, chances are that it's electronic parts will also be less reliable. Embedded hardware passes stricter quality checks than typical PC hardware, making it less sensitive to temperature and power supply anomalies.

Software that runs on embedded hardware, like CISCO IOS, tends to be more reliable as well, only because it is more simple in design. It has fewer layers, simpler interfaces and a smaller codebase, so there is smaller room for mistakes and debugging is easier in the event of them. From the administrator's standpoint, a router running on GNU/Linux will require more frequent maintenance to be reliable than an embedded router.

So if reliability is indeed a top priority, embedded routers are a better choice. However in my opinion reliability of this level is only necessary on very mission-critical environments, in most cases a possible hardware failure every couple of years and a few-minutes scheduled downtime every couple of months/weeks is not a problem.

Features. This is where a Linux-based router would win without doubt. Linux' rich features place it way ahead of anything else, I don't think there is much to say here: if you can imagine something -anything- and you know it's mechanics, you can 100% implement it on linux, and in more than one possible ways. Embedded solutions might offer sufficient features for most environments, but remain highly inflexible: if a custom need shows on the way, the associated features might not exist, their available implementations might be less than ideal for that particular case, or an expensive hardware upgrade/replacement may be required.

Knowledge requirements go hand to hand with the features. Linux is a generic operating system that gives the administrator complete control over every administration detail of the system's resources while the embedded routers' software only gives control over the very specific things outlined by their features. As a matter of fact, the producers of embedded routers make sure that their hardware can only be used for the specific things they marketed as features. It is easier to offer support this way, and it is more profitable to them, since they sell features "per piece". Linux developers on the other hand follow such an open design, that their software is possible to be used in ways that they might have never even thought about! This open design comes at the cost of knowledge requirements.

A Linux administrator must know many things in detail, regardless of whether he needs the advanced features associated with them. A CISCO administrator in most cases can have the luxury to simply ignore the features that he doesn't need and only refer to the documentation for the features that he needs, for which the knowledge requisites are limited. Any software that runs on Linux has a quite large chain of knowledge requisites and the user must be willing to follow that chain to it's end before he can use the software effectively (luckily, though, most of the links are common, so what he learns once applies everywhere). In environments where a new technology must be implemented in the minimum amount of time, that is unacceptable, especially if the available personnel is not really comfortable with the specifics of Linux.

Cost, of course, is an area that Linux has an advantage. It is completely free and can run on a wide variety of hardware, from x86 museum-pieces to high-end DEC/Compaq Alpha hardware.

Thank you very much!

You are indeed a person who knows what he is talking about, and mainly i agree ( i don't know linux that well...but i take your word for it )

The reason i ask ...I work at an ISP and as i said i recently got my CCNA. My boss wants me to continue with CCNP and then CCSP...i want to do this, but then i'm afraid i won't be able to learn linux and to know it as a pro.
i know linux but i can not do in linux what i'm doing in cisco IOS ( because i don't have the knowledge yet).So...i don't know what to do next...i think you will tell me to learn them both...but that's easier said than done

Regarding the features...i'm not that sure that in linux you can do a good BGP...i know about zebra, but i've heard from people that is not that good; on the other hand the cisco implementation of BGP is very good

Anyway thanks again for your reply!

Best regards,

G

Regarding the features...i'm not that sure that in linux you can do a good BGP...i know about zebra, but i've heard from people that is not that good; on the other hand the cisco implementation of BGP is very good

Zebra is since 2003 a dead project, most people use it's popular fork Quagga . There is also Xorp and OpenBGPd , however Xorp is too slow and OpenBGPd has not been ported properly from OpenBSD. For OpenBSD though, OpenBGPd is the best choice.

Quagga might perform worse than CISCO routers on commodity hardware, however it still performs well enough to support even a network of about 6000 nodes like the Athens Wireless Metropolitan Network.

In an ISP I don't think they would use Linux Routers, because cost is not so much of an issue, however they will definetelly use linux on their servers. If you like computers, play with Linux/*BSD in your free time and you will not regret it, neither personally nor professionally