From: Jongsoo kim (bstrt2002@gmail.com)
Date: Mon Apr 04 2005 - 03:25:14 GMT-3
Folks Thanks for all the excellent feedback .
Based on group's feedback and my trial test to see how pratical and
efficient my check list,
I revised some of them. Also I was advised that I can't bring the outside
pens so that I will do coloring with those color pencil available on desk.
#1 Spend a few minute to understand the point distribution between
Core requirement (L2, IGP, BGP, ISDN) and non-core ( IOS, Service,
Security, Mcast)
#2 Spend a few minute to understand the topology.
Figure out core network, stub network, BB
#3 Enter Alias command to notepad and copy paste all router.
"show run | b Se" ( surprizingly, I didn't use this command after I build
drawing because I can find out my sub-interface number from drawing!
#3 Attack F/R ( targetting 10~15 min)
While reading the task,, Draw a quick diagram showing interface type ( ph,
m, p2p).
Configure Router by router not interface by interface
Always 0) shut 1) enc frame-remay 2) no frame inverse 3) no shut.
Ping from spoke to spoke if possible. to vaildate.
If PPP over FR, then always create VT first, user/password
In this way, I was able to do this in 7 min for 3 pvc's ( each pvc has
different interface type).
#4 Attack CAT ( 25~35 min)
4-1While reading the task, make VLAN table like below
VL Router CAT1 CAT2 Router VL
10 R1 f0/0------f0/1 f0/2 ---------f0/0 R2 10
20 R3 f0/1------f0/3 f0/4 ---------f0/0 R4 30
40 R5 f0/0 ------f0/5
40 R6 f0/1-------f0/6
f0/23---f0/23
f0/24---f0/24
vl 10 vl40
client vtp server vtp
Determine VTP mode, trunk mode.
4-2 Delete vlan data base " delete flash:vlan" before configuring !
Then configue 1) VTP, 2) Vlan, 3) cat-cat 4) access port, 4) trunk port
4-3 Read task once again and make sure nothing missed
4-4 ping vlan by vlan. Select only one device and ping all other on a
specific vlan.
> No need to ping from multiple interface on a same vlan.
> Don't wait for Arp resolution!
CAT takes about 25 minutes in my scenario ( but real lab would take shorter)
>
> #5 Attack ATM ( I can spend a lot time if I screwed config. 5~25min )
> Quickly decide PVC vs SVC
> 5-1 If SVC, then decide "CLIP" or "SVC nsap"
> Put "pvc 0/16 ilmi and pvc 0/5 qsaal " and "show atm ilmi-status" to
> vaildate nsap address.
> 5-1-1 if CLIP, then decide "arp-server self" or "arp-server nsap"
> And then decide physical or sub
> 5-1-2 if SVC nsap, decide physical or logical
> 5-2 if PVC, then decide "pvc vci/vpi" or map-list/map-group
> 5-3 after 5-1 or 5-2 done, figure our nsap or vci/vpi. Pay attention
> nssp is HEX!
If PPP over ATM, then always create VT or dialer interface first, then
> user/password
> 5-4 ping and validate
############## L2 is over between 40~1:15
###########################################
>
> #6 Attack OSPF
Based on my test, it was very important the way I write down on paper in
order to make router-by-router step work.
6-1 While reading the task, Draw a diagram to configure OSPF router by
router not area by area w/ green coloring.( 10 min)
> Check if there are
> authentication
> stub or nssa.
> virtual link
> Make a note on redistribute, summary, area-range.
> Pay attention DR/BDR, OPSF network type
Write config separately for interface and ospf on drawing.
For example, the below was my note on drawing I made while I am reading
task.
For R1
int s0/0.123
p2m, md5,
int s0/0.14
non-bro, pri 0, md5
int f0/0
nothing
ospf
r-id
a 0 md5
a 12 nssa no-sum, no-red
a 13 stub no-sum
a 12 v r4 md5
a 14 v r2 md5
a 5 v r3 md5
nei R2
neii R3
area 5 range
summary
This method makes configuration time very short but it was extremely
important to not forget anything to configure router-by-router as many
people pointed out.
> 6-2 Configure OSPF router by router based on drawing in Black ( 10~30 min)
First Interface and then router ospf
> 6-2-1 Preferred sequence for configuring interface was 1)OPSF network type
based, 2) priority, 3) Authentication,
> 6-2-2 Preferred sequence for configuring OSPF process was from
"easy-to-forget" to "always know" 1) router-id( it seem to only help for
Virtual-link, I will skip if there is no Virtual link ) 2) area
authentication, 3) area virtual link, 4) neighbor, 5) Network (copy past
from interface address)
> 6-2-3 Validate everything is working( show ip os ne, show ip os vir, show
ip os interface, show ip route ), ( 5 min)
Just a note wth this method, I was able to do OSPF among five routers in 15
min from drawing to core config not including redistribute/summary/area
range. This is my record time.
Specially, virtual link config really seems to save time.
There isn't much of trap in OSPF like Rip. very easy to validate it as well.
If config work, in most case it should be correct...
6-3 Do redistribute, summary, area range ( 5 min)
It was necessary to separately treat area range, or summary
6-4 avoid any engagement with giant beasts. But make a note.
#####OSPF is from 35 ~ 55 Min ( total 1:15 ~2:10)#######
> 7 Attack RIP( 20~30 min)
> It is very tricky!
> 7-1 add RIP topology into OPSF drawing with blue coloring( 2 min).
It seems Rip doesn't really have much detail info on drawing.
> 7-2 Make sure active/passive interface
WATCH OUT Split-Horizon is off on pfysical FR and ATM !
> Pay attention of rip update method ( M/B/U) and version, authentication
> Never assume it is always V2!, no auto-summary, mcast, etc
> This selection can be applied to each direction of interface.
> 7-3 Configure router by router( 5 min) per drawing
In fact, core rip configuration is always router by router as rip doesn't
have concept of adjacency per link.
> 7-4 valiadte ( 3 min)
> 7-5 Spend enough time to be absolutely correct on route-filter,
> summary, etc ( 5 min)
> 7-6 If mutual-redistribution is required, make sure multi-exit point
> ot single-exit point. Don't fotget metric.
> If it is multi-exit point, write down "rip subnets" on notepad and do
> the following( 5 min)
> 7-6-1 "redistribute ospf" under "router rip"
> ##### Provent Rip-originated routes entering Rip from OSPF ############
> "Deny rip routes and permit all" route-map for "redistribute ospf" to rip
> Don't wait after "clear ip route * " is issued if I am not "idiot!"
>
> 7-6-2 "redistribute rip subnets" under "router ospf"
> ##### Provent OSPF external routes entering OSPF from Rip #####
> "Permit only rip routes" route-map for "redistribute rip subnets" to OSPF
> Don't wait after "clear ip route * " is issued if I am not "idiot!"
>
> 7-6-3 distance 121 0.0.0.0 <http://0.0.0.0>
255.255.255.255<http://255.255.255.255>11 under "router OSPF"
> ##### Fix redistributing router's AD for Rip routes #####
> distance 121 0.0.0.0 <http://0.0.0.0>
255.255.255.255<http://255.255.255.255>11
> "access-list 11 permit rip routes"
> I saw sometimes this takes quite a few second. Don't do "clear ip
> OPSF" or I will end up spending more time just for watching.
>
#### RIP is over 20 ~30 min( total 1:35 ~ 2:40) ############
>
> 8 Attack EIGRP ( 20~30min)
> 8-1 While reading the task, add EIGRP topology into OPSF drawing in black
w/o blue coloring ( 2 min).
> 8-2 Determine non/passive/active-eigrp interface. Be open minded that
WATCH OUT Split-Horizon is off on pfysical FR and ATM !
> BB can be multicast/unicast. Load-balance, authentication, stub,
> summary address( 5 min )
> 8-3 Configure router by router( 5 min) per drawing
> 8-4 validate ( 5 min)
> 8-5 Spend enough time to be absolutely correct on route-filter,
> summary, etc ( 5 min)
> 8-6 If mutual-redistribution is required, make sure multi-exit point
> ot single-exit point.
>
> If it is multi-exit point, write down "eigrp subnets" on notepad ( 5 min)
> 8-6-1"redistribute ospf" under "router eigrp"
> #####Protect EIGRP external route reentering from OSPF #######
> "Deny eigrp routes and permit all" route-map for "redistribute ospf" to
eigrp
> Make sure metric is configured.
>
> 8-6-2 "redistribute eigrp subnet" under "router ospf"
> ##### Protect OSPF external routes reentering from EIGRP
> "Only permit eigrp routes" route-map for "redistribute ospf" to eigrp
> Make sure metric is configured.
>
> 8-6-3 distance 121 0.0.0.0 <http://0.0.0.0>
255.255.255.255<http://255.255.255.255>11 under "router OSPF"
> ##### Fix redistributing router's AD for eigrp external routes #####
> distance 121 0.0.0.0 <http://0.0.0.0>
255.255.255.255<http://255.255.255.255>11
> "access-list 11 permit eigrp routes"
> I saw sometimes this takes quite a few second. Don't do "clear ip
> OPSF" or I will end up spending more time just for watching.
> Technically, only eigrp external route needs to be applied but eigrp
> route won't hurt and make it simple.
>
######EIGRP is over in 20~30 min (1:55 ~3:10 min) ##############
>
> 9.Attack ISIS ( 10 min)
> 9-1 While reading the task, add ISIS topology into OPSF drawing in black
w/ purple coloring ( 1 min).
> 9-2 determine area type, IS-type, authentication ( domain, area,
> interface level1-2).
> Make sure of correct value of NET ( it is Hex), summary address
> 9-3 Configure router by router.
> 9-4 I don't believe there will be multi-exit mutual redistribution on ISIS
> Make sure to redistribute connect network from ISIS to OSPF.
>
###### ISIS is over in 10~15 min ( 2:05 ~3:25)
>
> 10 Attack ISDN ( 15~30 min)
> 10-1 draw ISDN on a separate paper. ( 30 sec)
> 10-2 Determine single/both callers, authentication type( no
> auth/pap/chap), physical/dialer interface. PPP feature = multilink,
> callback,
> 10-3 Figure out back-up method ( floating static/OSPF demand/watch
> group/back-up interface/rip trriger/ snap-shot routing ) focus on how
> full reachability can be accomplished after F/R failed. Make sure
> link is not flapping.
> 10-4 Determine if there is additional task for interesting traffic
filtering.
> 10-5 configure ISDN router by router.
> 10-5-1 select switch type, spid and shut and no shut and show isdn status.
> make sure L2 is happy! Also make a quick test call using both
> string " isdn test call interface bri0/0 "string" " and disconnect "
> isdn test disconnect interface bri0/0 all"
> 10-5-2 validate the link
>
###### ISDN is over in 15 ~30 min ( 2:20 ~ 3:55)
>
> 11 Golden Moment ( 5~30 min)
11.1 Test full reachability with ISDN back-up link off
> Check the Golden moment per NMC meaning the exciting moment when you
> get ping response from every router to every router.
> Run tclsh script
> "foreach addr {
> 1.1.1.1 <http://1.1.1.1>
> ...
> } { ping $ addr}"
> Just copy past after tclsh ( it is really cool when you see pings go
> through from everywhere to everywhere). To quit, juts type " tclq"
11.1 Test full reachability with ISDN back-up link on
11.2 when ping has no response, write down ip address and troubleshoot.
> Drawing will be the excellent tool for troubleshooting
> Don't bother ISDN link yet.
>
########### Full reachability is done in 5 ~30 min ( 2:25 ~4:25)
> 12 Attack BGP( 20 ~40 min)
> 12.1 While reading task, Drawing a BGP topology on a separate paper.( 3
min)
Drawing is very imnportant in BGP
12.2 Determine RR or CON or both to do full-mesh iBGP.
> See if neighbor peer-group is required,
> decide ip address ot use bgp session.
> 12.3 Configure router by router not BGP session-by-session
> always put no sync and no auto-summary if allowed.
> 12-4 Spend enough time to be absolutely correct on route-filtering (
> ACL, prefix-list, as-path filer), route-aggregate(w/ as-set,
> summary-only, supress-map, attribute-map, advertise-map),
> route-manipulation( w/as-prepending, med, local-pref, weight,
> next-hop, advertise-map/non/existing-map, orgin, community, etc )
> route-dampening, etc.
> 12-5 vaildate config. Use "clear ip bgp * soft " not " clear ip bgp * and
I don't have to wait!
>
###### BGP is over in 20 ~40 ( 2:45 ~ 5:05) My target is before lunch!
>
> 13 IPv6( 10 min)
> 13-1 draw a sipmple diagram ( 1 min)
> 13-2 Watch out link local address over FR multilink.
> SLA ID is 4th 16bit
> 16bit:16bit:16bit:SLA ID(16 bit) : interface ID( 64 bits)
> site-local = FEC0::
> link-local = fe80::
> 13-3 Check a full reachability using tcl script or just manual ping
> depneding on the number router.
>
> IPv6 is over 10 min ( total 2:45 ~ 5:15)
>
> ################## Core routing is done ####################
> I should have at least 2:45 hours to go at least.
>
> Strategy will change depending how much time I have at this moment.
>
> 14 I would do multicast first ( 15 min)
> 14-1 While reading task, mark a Mcast topology with red high lighter on
OSPF drawing.
> 14-2 Determine mcast topology ( dense-mode, static RP pim sparse,
> Auto-rp/MA, pim V2 bsr, Auto-rp/MA/MSDP).
Spot any RFP issue per IGP topology
> 14-3 Configure router-by-router
> 14-4 valildate it
> 14-5 If second part is difficult, skip by making a note.
#####Minimum 2:30 left
> 15 IOS/IP service ( 25 min)
> Be careful not to block or drop any IGP updates
> 15-1, just check quikcly and do easy one first.
> 15-2, skip difficult task by making a note
###### minimum 2:05 left
> 16 QoS ( 30 ~ 40min)
> Be careful not to block or drop any IGP updates
> 16-1 Draw a flow on paper instead of in brain.
> 16-2 Always determine classification method( ACL, NBAR) and direction.
> 16-3 Determine shaping vs policing
> 16-4 Consider all options for queuing( legacy custom/priority,
> bandwidth/priority, shape average/peak, FRTS/GTS)
> 16-5 consider all options for policing ( police, rate-limit, ip
> multicast rate-limit, aggregate police( 3550))
> 16-6 If frame-relay, don't forget adaptive-shaping.( becn, fecn,
foresight)
> 16-7 Consider all droping mode (random detect, ecn, tail drop, marking,
etc)
##### minimum 1:25
> 17 Security ( 30~40min)
> Be careful not to block or drop any IGP updates
> 17-1 Draw a flow on paper instead of in brain.
> 17-2 Consdier all options for classification
> std/ext/reflexive/dynamic ACL,
> IP insepct,
> tcp intercept
> unicast RFP,
> ip accouting output packet /access-violation/precedence,
> 17-2 When configuring Switchport port-security mac-address, be careful
> to include vurtual and physical mac if HSRP is running.
###### minimum 45 min
> 18 DLSW( 15 min)
> 18.1 Draw a qucik topology ( 1 min)
> 18.2 Decide method of DLSW TCP, fst, fr.( I think only TCP will show up)
> Peer on-demand( group/border)
> Dynamic peering ( dynamic)
> Loadbalance (round-robin, circuit-count),
> Back-up ( back-up peer or cost)
> DSLW use tcp 2065 and udp 2067
> NAT can affect DLSW ( higher ip DLSW peer drops)
> 18.3 decide type of filtering
> 18-3-1 Netbios name filter( netbios access-list host xyz permit zyx )
> Icanreach/icannotreach netbios-name /netbiosexclusive
>
> 18-3-2 MAC address filer ( access-list 700-799, mac-address conevrsion
needed )
> Icanreach/icannotreach mac-address/mac-exclusive( address conversion)
>
> 18-3-3 LSAP filter ( access-list 200-299 permit )
> SNA only "access-list 200 permit 0x0000 0x0d0d"
> SNA and Netbios " access-list 200 permit 0xf0f0 0x0101
> Icanreach/icannotreach saps
> icannotreach saps f0 ( deny netbios)
>
###### minimum 30 min #############
I am planing at least 1:30 hour left.
I will do " tcl script " one more time to make sure everything work.
I expect 2 ~ 4 question I will skip.
At this moment, depending on how much time I have, I quckily go back to the
qeustion I skipped.
I will invest my time to something I can see best chance of getting right
out of the skipped ones.
Jongsoo from RTP
This archive was generated by hypermail 2.1.4 : Tue May 03 2005 - 07:54:52 GMT-3