6509 used as non Core devices

routingbyrumor

Hi! Can someone clear this up for me I am quite confused. The 6509 is meant to be a core device since its so expensive and does switching and routing at fast speeds. However, in my organization they are using two 6509's as core devices and then cascading another two 6509's together providing access to workstations using blades. Is this correct to do? Shouldn't the 6509's be reserved for the core and less expensive switches be reserved as an access layer device for workstations? If so then why do they make 48 port blades for the 6509's?

Thanks.

Humper

Well our college here uses 6509's throughout the entire school. Generally yes, the 6500s are used for cores but if you have the money chassis are much better!

mikej412

Port density, backplane speed, upgradability -- and Voice support.

keenon

well for me ..

this is my opinion: 6500s should be used at the core or distribution layers of the network .. servers i would use a 6500


for terminating workstations not if i have a say so..

mikej412

Oh -- and the product information (a.k.a. marketing stuff) should have at least 1/2-bazillion reasons why everyone needs one in their living room.

According to the diagram on that marketing page -- the 6500 is at home in the distribution/core, data center access....
and <drum roll> wiring closet.

Oh -- High Availability. One highly available device with redundant components is easier to manage than lots of smaller devices, has fewer points of failure, and greater uptime. You can have extra switches stacked in the wiring closet waiting to be swapped in -- but you lose uptime.

routingbyrumor

Thanks for the help guys I am trying to decide if I can use this against them and say that the 6509 should be strictly a core/destribution device.

keenon

the sheer cost alone is killer.. the chassis has its own price.. not to mention adding sup(s), blades, power supplies $$$$$$$$$


try pricing 1

mikej412

keenon wrote:

try pricing 1

I was saving this reason -- could be number 1 if this were a Letterman top ten list.

And the number one reason for finding 65xx routers in a wiring closet when there is no technical reason (even including supporting real business requirements) -- the sales rep with a "skybox" for sporting events.... and a non-technical IT Manager with a large budget who likes sports....

routingbyrumor

mikej412 and keenon,

So what would you say about an organization that is currently doing this? Would it be a waste of money, should they redesign the network? I am currently assessng the network and feel that they should migrate these devices to the core of the network rather than serve them as access layer devices. But since they already have such a large amount of users attached to these switches would this be a worthy mention?

Thanks.

Humper

If they can afford/have chassis at the access layer then I wouldnt change it.

PCHoldmann

I agree with ModemHumper, if they are in place, and there is no compelling reason to deploy them elsewhere, I see no reason to take them out.

There are reasons that you may want to use 6500 series switches like this, including the increased backplane speed, better VoIP support, availability, etc. The main downside is and always will be the cost. If the straight per-port cost was the same, I see no reason to deploy them anywhere, and if there is a reason to pay the higher per-port cost, I don't see it doing anyone (except maybe shareholders) any harm.

Humper

I believe a 4510 or 09 with some blades costs around $120,000. Depends on what you want to purchase.

keenon

4510 run about 75k+ 1 sup, 2PS( depending on the type the cost will vary), and 5 copper blades(7- 12k) .. i have installed quiet a few of these lately in places i wouldn't want to but had to ... 6500s depending on the model 06, 09, 13 can hit pretty high. 6513 with a sup 720 (25k), 3- 48 port blades ( like 10-15k each) is easily 100k+ really closer to 120k with the upgraded memory. chassis has its own cost 5 - 8k

i dunno.. i still prefer using access types switches in the closets and invest heavier at the distribution and/or core.

mikej412

routingbyrumor wrote:

So what would you say about an organization that is currently doing this? Would it be a waste of money,

I mentioned reasons why you might do it.... And the skybox (or sports tickets, or free golf, or free trips on corporate jets to trade shows in Las Vegas) may all be reasons they were purchased and were a waste of money.

Have you asked someone in your organization about the business/technical requirements that led to the decision to purchase those 65xxs? Are the users attached performing any mission critical functions? What's the cost per user of network downtime? Any contractual obligations that could be compromised by replacing the 65xx with "cheaper" switches. Any other modules other than switchports?

If no one is talking (or isn't there anymore), what is the configuration? bandwidth? #ports? #users? #servers? Look at the configuration (or get someone to printout a "show tech-support"). Are you a large organization with high density requirements (from that first marketing link I posted)?

If you're not sure what you're looking for (or need to ask) -- look at the Cisco Catalyst 6500 Series Switches Introduction page. Are you doing any of the things mentioned there that the 6500 does?

I mentioned the high availability, Voice, redundancy, etc already, but totally forgot the security reasons.

routingbyrumor wrote:

should they redesign the network?

If it ain't broke -- don't fix it. Unless there is a need for another core router (or a hot spare) -- if you pull it out, you'll be spending more money on equipment to do the job the 65xx is already doing.

routingbyrumor wrote:

I am currently assessng the network and feel that they should migrate these devices to the core of the network rather than serve them as access layer devices.

Cisco says they are not just core routers -- what are your reasons for putting them in the core -- other than you think they should be core routers?

routingbyrumor wrote:

But since they already have such a large amount of users attached to these switches would this be a worthy mention?

If you do it during non-business hours, it would only piss off the people who had to do the swapout/install (unless they are already there 2nd/3rd shift and bored and hoping for some work excitement)

routingbyrumor

Thanks guys you have given me more insite into the situation. The company is actually a farily small institute providing education. I acutally go to school there and am accessing there network readiness for MPLS. Our senior project job was to give recommendations on how they could improve the network. They currently have 4 6509's in the network along with a mirad of other lower end Cisco switches. There is no need for high availibility but it would be nice. There is no VoIP either. I wanted to recommend the 6509's as core devices because they are currently termininating all connections to the first floor for all lower end switches as well as the large 6500's. (single point of failure) My recommendation involved turning the current 2 6509's on the third floor into core devices and implement HSRP between the 2 6509's on the 3rd floor and the 2 in the current core. (located on the first floor). They could then terminate the smaller end switches on the third floor and terminate some other smaller switches on the first floor at the core. This way it would be more resilient to failure as well as save on cost from running fiber throughout the building.

mikej412

routingbyrumor wrote:

My recommendation involved turning the current 2 6509's on the third floor into core devices and implement HSRP between the 2 6509's on the 3rd floor and the 2 in the current core.

If 2 of the 6500s are already the core, and you have dual connections to the other 6500s, and are using Layer 3 -- you already have a "dual-path L3 core design" and should have two equal cost paths to every network destination. Running HSRP basically puts 1/2 your available bandwidth in "standby mode."


Are you sure you're not already configured as a collapsed backbone or combined distribution/core layers? Since you said you're not "large" -- have you confirmed whether the current core is layer 2 or 3?

For MPLS support -- check the feature navigator to make sure your current IOS feature set has support for MPLS.

routingbyrumor

Thanks, Mike I'll take your advice.