[c-nsp] Cisco core router (for smaller sized colocation provider) recomendations please

Nick Shah Nick.Shah at aapt.com.au
Mon Jan 23 18:51:42 EST 2006


Josh

For past 5-7 years, we have successfully (business wise) run 3 types of
data centre models.

- a low-budget 'el cheapo' data centre model, consisting of 7206 (NPE200
to begin with and now all of them NPE-G1) with a trunk connecting to
2948's (and more latest onto 3550/3750) offering customers 1 x 10 or
100Mbps speeds & rack space. If real estate is not premium priced (we
own most), this model had the potential to make some good mullah. To
make up for redundancy we have multiple switches & multiple 7200's in
data center, and customers wanting redundancy buy feeds of 2 or more
switches. 

- To tackle the growth and deliver speeds upto 1GIG, we moved a 6509
model, again simple single VLAN per customer (no real complexities).

- We also have a premium offering where we built a dual GSR core
connecting to dual 6509 distribution, connecting to 2948G's access
offering multiple redundant feeds to customer racks. This model utilised
PVLAN on the access (1 x PVLAN per customer)

So, in my opinion if you were to build an el-cheapo (pun unintended)
offering multiple 7206 NPE-G1 (or even a 7301s) trunked to multiple
switches (build your own redundancy) offering redundancy at layer 2 or 3
should be quite economical, but wont scale (scalability here would mean
throwing more boxes at it).

A nice little budget, yet fully redundant and scalable offering would
consist of dual 6500's, and use gigabit switches (3750's) at the edge. 

One of the key factors would be whether you want to build redundancy at
layer 2 (and offer it to customers) which would be expensive for you to
build, the other end of the spectrum is 'trunking on a stick' 7206 -
3750 model where customer has to buy redundancy if they need it. 

Ps. I may be saying things you already know, and possibly not making
your decision any easier, but it ain't an easy decision and that's why
we are running 'horses for courses' and still sticking to 3 x different
data centre designs.

Rgds
nick

Nick Shah
CCIE# 10474
-------------------------------
AAPT Ltd.
180 Burnley St.
Richmond VIC 3121
 
Ph: +61 3 8687 5757
Mo: +61 414 825 956
-----Original Message-----
From: cisco-nsp-bounces at puck.nether.net
[mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of josh harrington
Sent: Tuesday, 24 January 2006 10:12 a.m.
To: cisco-nsp at puck.nether.net
Subject: [c-nsp] Cisco core router (for smaller sized colocation
provider) recomendations please

Hello, hope this isn't too far offtopic here but being a troller for a
long
time here I've realized there is a great knowledge base so I wanted to
at
least see if i could get some tips.  I help run a small colocation
company
in California and I am in the middle of recommending a new 'core router'
platform for our network.  We offer mainly colo and dedicated servers,
and
several of our clients use our space for VOIP services so quality even
under
high peak usage is a must.  We are not huge, but as we have had near
200%
growth in the past 12 months and need to expand our network asap to keep
up.
Simply put, I'd love to hear feedback and/or suggestions from any of you
guys who have gone through this already.

Our network map is real simple:

[Carrier 7609] --> 100 mbit --> Our cisco 7206 --> 100 mbit --> racks

[the racks on our end are a series of switches, mainly 2948gl3's]

We push about 60 mbit to/from our (1) carrier at peak right now, and the
router keeps up fine [its a cisco 7206 npe 150 btw, very low end on the
7206
line], and at peak we have under 50,000 packets per second, and our 7206
has little/no features enabled [just static routes and passing all
traffic
between 2 Ethernet 100 mbit interfaces].

To date we have had 2 problems, both were DOS attacks launched FROM one
of
our customer's servers flooding a full 100 mbit wire with more packets
per
second than the router could handle (the 2948gl3's spiked to about 50%
cpu
load during the attack but the 7200 literally just died for 3 minutes as
the
interface(s) all rebooted].  So our main goal to grow is something that
can
handle a lot more in this arena against a DOS, and handle our future
growth.

In then next 12 months we plan to add a 2nd carrier, at t3, 100mbit, or
possibly oc3 speed, and possibly upgrade our main carrier to a GigE
connection.  Probably maxing combined in the 300 mbit range, more likely
closer to half that in 12 months.

==== Problems/Requirements ====
- Budget is in the $5k to $20k range ($20k if its going to outlast me
even
past my 12 month projections)
- must not 'collapse' under simple packet flow DOS attack
- must handle BGP4 from 2 carriers with full route tables
- We plan to buy used, prices below are based on USED, 30 day warranty
ebay 
postings

===== Choices/Options that we have looked at: ====
Option #1: Cisco VXR 7206 [$4k to $12k]
Option #2: Cisco 12008 [$7k to $14k]
Option #3: Cisco 6509 [$10k to $15k]

Here are the 3 main options, broken down a bit more in depth. [I have
not
ruled out juniper all together, but not enough experience with them and
lots of experience with cisco, makes cisco our better option i think,
especially since its easier to find used cisco gear than it is to find
used
juniper gear at a decent price].

[option #1 - Cisco 7206 VXR]
--------------------------------
Estimated: $4,000 [$6,000 with 400 mhz, $12,000 with the 1 ghz cpu
upgrade]
1 Cisco 7206 VXR NPE 300 mhz w/max ram
2 AC Power
2 Fast Ethernet Adapters (1 included on the NPE)

+ lots of experience on this unit
+ lots of spare cards (most compatible)
+ can keep old 7200 as a hot standby, minimizing long term downtime
- END OF LIFE/sale/support on most of the 7200 product line over 5 years

ago! The VXR model is darn close to end of life i suspect
- minimal horse power here for the money, prone to death by packet
attack

[option #2 - Cisco GSR (12008)]
--------------------------------
Estimated: $7,000 to $14,000 [varies if I start with GigE or just
100mbit]
1 Cisco12008 GSR 40Gbps
1 Clock Scheduler Card (GSR8)
3 Switch Fabric Card (GSR8)
2 AC Power
1 4 port OC-3c/STM-1 Single Mode
1 GE card or a 4 port x 100 mbit

+ much higher total bandwidth/packet processing power compared to 7200,
for
similar money
- product is long since obsolete and outclassed by the 760x cisco
router, as
well as just about any juniper router in the m20+ tier.
- I'd bet if i buy this, cisco will classify it end of life within 3
months 
:)
- over priced 'blade cards' to add any other functions/circuits (high
costs)

[option #3 - Cisco 6509 switch'router' w/MSFC2]
------------------------------------------------------------
Estimated: $10,000 - $15,000 (and up depending on config)
1 WS-C6509 Cisco Catalyst 6500 9-Slot Chassis
1 WS-C6K-9SLOT-FAN Catalyst 6000 Fan Tray for 9-Slot Systems
1 WS-C6X09-RACK Catalyst 6x09 Rack Mount Kit
2 WS-CAC-1300W 1300W AC Power Supply
1 CAB-7513AC AC Power Cord
1 WS-X6K-S1A-MSFC2 Catalyst 6500 Supervisor Engine-2, 2GE, plus MSFC-2 /
PFC
(WS-X6K-S1A-2GE + MSFC-2 & PFC)
1 MEM-C6K-FLC24M  24MB Flash Card
1 WS-X6408A-GBIC Catalyst 6500 8-Port Gigabit Ethernet Module (Req.
GBICs)
1 WS-X6348-RJ-45 Catalyst 6500 48-Port 10/100 RJ-45 Module


+ could consolidate router/switch into 1 [i.e. replace my 48 port
switches 
in each rack]
+ still a major product with sales/support, no end of life 'soon', this
is 
still a primary flag ship product
- 'not a router' as some would say [though this one is as good as it
gets
for a switch with router ability built in, so i read at least]
- bgp4 support appears limited in previous versions, but the MSFC2
processor
supposedly can handle (2) bgp4 sessions properly [makes me nervous]
- no support for anything but 100mbit, or gigE links, wont work with t3,
or
oc3 lines [since i don't know what ill buy from my next carrier this is
a
draw back since i may very well get a circuit this switch/router can't
use]
- 'all eggs in 1 basket' theory, if it breaks you loose all your ether
switches! [at least with separate routers/switches i can swap in an old
7206
router spare and get back online fast in a worst case scenario.


-- OK so if you read this far, you probably wonder why the heck am i
asking
you guys and not calling cisco and juniper, and others?  Well I have,
and I
have, they all gave me their input- but I'd love to hear from some real
people in the field rather than sales engineers and bloated brochures
claiming much more capacity than true world usage will produce etc.

                          - Josh Harington  jharington68 at hotmail.com

_________________________________________________________________
On the road to retirement? Check out MSN Life Events for advice on how
to 
get there! http://lifeevents.msn.com/category.aspx?cid=Retirement

_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


This communication, including any attachments, is confidential. If you are not the intended recipient, you should not read it - please contact me immediately, destroy it, and do not copy or use any part of this communication or disclose anything about it.





More information about the cisco-nsp mailing list