Re: [c-nsp] Cisco ASR vs Juniper

[Mark Tinka]

On 5/24/17 4:35 PM, Aaron Gould wrote:

> About the MX104 and ACX5000
>
> I have ~7,000 dsl customers being nat'd behind /24 of address space on a
> pair of MX104's... they run nicely on two mpls l3vpn's... nat inside vrf
> (ri) and nat outside vrf (ri)

The RE sucks. It's too slow.

We are now running them out in some sites and replacing them with
MX480's. Pity, since the MX104 data plane is solid.


>
> I have deployed (~30) ACX5048's as mpls p's and pe's and they are running
> well.  I have hit a bug with VPLS that requires a vpls routing-instance
> bounce to revive, but JTAC just told me the PR is hitting D20 software and
> fixed in D25 still need to test that.  But all in all, I like the
> ACX5048's.

For our use-case, the Broadcom in those routers presents some
limitations Juniper are never going to fix.

That said, the Broadcom chipset does make them cheaper, but you also pay
for that in other ways.

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

Re: [c-nsp] Cisco ASR vs Juniper

[Aaron Gould]

...i re-read some of your criteria... ummm, so I use MX104's and ACX5048's
with MP-iBGP for just learning my internal core routes, not big table for
world routes... so for what I use those boxes for, they are nice.

-Aaron

-Original Message-
From: cisco-nsp [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of
Aaron Gould
Sent: Wednesday, May 24, 2017 9:36 AM
To: 'Mark Tinka' <mark.ti...@seacom.mu>; 'Mark Mason'
<mma...@jackhenry.com>; cisco-nsp@puck.nether.net
Subject: Re: [c-nsp] Cisco ASR vs Juniper

About the MX104 and ACX5000

I have ~7,000 dsl customers being nat'd behind /24 of address space on a
pair of MX104's... they run nicely on two mpls l3vpn's... nat inside vrf
(ri) and nat outside vrf (ri)

I have deployed (~30) ACX5048's as mpls p's and pe's and they are running
well.  I have hit a bug with VPLS that requires a vpls routing-instance
bounce to revive, but JTAC just told me the PR is hitting D20 software and
fixed in D25 still need to test that.  But all in all, I like the
ACX5048's.

-Aaron

-Original Message-
From: cisco-nsp [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Mark
Tinka
Sent: Wednesday, May 24, 2017 2:16 AM
To: Mark Mason <mma...@jackhenry.com>; cisco-nsp@puck.nether.net
Subject: Re: [c-nsp] Cisco ASR vs Juniper



On 5/9/17 7:29 PM, Mark Mason wrote:

> Alright crowd...Ready the rifles and prepare for battle...Cisco ASR or
Juniper. Cost, operability, chassis lifespan new vs. old, memory
requirements, etc. So many details. Feel free to take the post anywhere
you'd like.

I'm really liking the new ASR1000 family of routers.

But we did the month since December last year, and any way we cut it, the
MX480 works out cheaper.

Stay away from the MX104 or ACX5000.

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

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

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

Re: [c-nsp] Cisco ASR vs Juniper

[Aaron Gould]

About the MX104 and ACX5000

I have ~7,000 dsl customers being nat'd behind /24 of address space on a
pair of MX104's... they run nicely on two mpls l3vpn's... nat inside vrf
(ri) and nat outside vrf (ri)

I have deployed (~30) ACX5048's as mpls p's and pe's and they are running
well.  I have hit a bug with VPLS that requires a vpls routing-instance
bounce to revive, but JTAC just told me the PR is hitting D20 software and
fixed in D25 still need to test that.  But all in all, I like the
ACX5048's.

-Aaron

-Original Message-
From: cisco-nsp [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Mark
Tinka
Sent: Wednesday, May 24, 2017 2:16 AM
To: Mark Mason <mma...@jackhenry.com>; cisco-nsp@puck.nether.net
Subject: Re: [c-nsp] Cisco ASR vs Juniper



On 5/9/17 7:29 PM, Mark Mason wrote:

> Alright crowd...Ready the rifles and prepare for battle...Cisco ASR or
Juniper. Cost, operability, chassis lifespan new vs. old, memory
requirements, etc. So many details. Feel free to take the post anywhere
you'd like.

I'm really liking the new ASR1000 family of routers.

But we did the month since December last year, and any way we cut it, the
MX480 works out cheaper.

Stay away from the MX104 or ACX5000.

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

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

Re: [c-nsp] Cisco ASR vs Juniper

[adamv0025]

> Mark Mason
> Sent: Tuesday, May 09, 2017 6:29 PM
> 
> Alright crowd...Ready the rifles and prepare for battle...Cisco ASR or
Juniper.
> Cost, operability, chassis lifespan new vs. old, memory requirements, etc.
So
> many details. Feel free to take the post anywhere you'd like.
> 
> Deployments:
> EDGE
> Pure internet router/edge with single eBGP toward ISP - 10Gb up to ISP and
> 10Gb down to Aggregation Point
> MX10 w/ licensing get me 4x10Gb (but sys cap. At 20Gb?), MX40 vs ASR1001X
> (or ASR1001-HX)
> +pretty vanilla installation
> +not running security at this level
> +just a router doing routing
> +single eBGP so/so on memory requirements single iBGP down to Core/Agg
> 
If it's just pure ipv4 Internet routing (i.e. no need for protecting high
priority packets) you might be ok even with 1st gen Trio from Juniper so
either one would do I guess.  

> AGGREGATION
> Internet Edge Aggregation/multiple iBGP sessions toward edge routers
> MX104 or MX240 (or greater chassis lineup) vs ASR9k
> +HUGE MEMORY requirements
> +Multiple BGP aggregation feeds
> +Installation of best routes from MULTIPLE carriers
> 
Yeah smells like big tables and long FIB download times.
You might want to use BGP PIC to speed things up here and be part of the
solution, but unfortunately Junos doesn't support BGP PIC for pure IPv4. 
MX240+ and ASR9k are modular so the comparison really depends on which
line-cards you're considering for each platform. 
But in general you might want to stay away from cards using first generation
of Trio chipsets and Gen1.5 Trio voodoo. And the same applies to first gen
(Trident) ASR9k LCs.   
XR has so much better BGP implementation than Junos. 


adam

netconsultings.com
::carrier-class solutions for the telecommunications industry::

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

Re: [c-nsp] Cisco ASR vs Juniper

[Mark Tinka]

On 5/9/17 7:29 PM, Mark Mason wrote:

> Alright crowd...Ready the rifles and prepare for battle...Cisco ASR or 
> Juniper. Cost, operability, chassis lifespan new vs. old, memory 
> requirements, etc. So many details. Feel free to take the post anywhere you'd 
> like.

I'm really liking the new ASR1000 family of routers.

But we did the month since December last year, and any way we cut it,
the MX480 works out cheaper.

Stay away from the MX104 or ACX5000.

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

Re: [c-nsp] Cisco ASR vs Juniper

[Mark Mason]

*bump* was never posted.

From: Mark Mason
Sent: Tuesday, May 9, 2017 12:29 PM
To: 'cisco-nsp@puck.nether.net' 
Subject: Cisco ASR vs Juniper

Alright crowd...Ready the rifles and prepare for battle...Cisco ASR or Juniper. 
Cost, operability, chassis lifespan new vs. old, memory requirements, etc. So 
many details. Feel free to take the post anywhere you'd like.

Deployments:
EDGE
Pure internet router/edge with single eBGP toward ISP - 10Gb up to ISP and 10Gb 
down to Aggregation Point
MX10 w/ licensing get me 4x10Gb (but sys cap. At 20Gb?), MX40 vs ASR1001X (or 
ASR1001-HX)
+pretty vanilla installation
+not running security at this level
+just a router doing routing
+single eBGP so/so on memory requirements
+single iBGP down to Core/Agg

AGGREGATION
Internet Edge Aggregation/multiple iBGP sessions toward edge routers
MX104 or MX240 (or greater chassis lineup) vs ASR9k
+HUGE MEMORY requirements
+Multiple BGP aggregation feeds
+Installation of best routes from MULTIPLE carriers
NOTICE: This electronic mail message and any files transmitted with it are 
intended
exclusively for the individual or entity to which it is addressed. The message, 
together with any attachment, may contain confidential and/or privileged 
information.
Any unauthorized review, use, printing, saving, copying, disclosure or 
distribution 
is strictly prohibited. If you have received this message in error, please 
immediately advise the sender by reply email and delete all copies.
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Re: [c-nsp] Cisco ASR vs Juniper

[Aaron Gould]

Hi James, I haven't done much with QoS on the ACX5048 yet.  When I do, I don't 
think we will be doing as much as you described with the ME3600.  I hardly did 
any QoS with my ME3600's occasional policer or shaper on efp here and there.

-Aaron

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

Re: [c-nsp] Cisco ASR vs Juniper

[Mark Tees]

From memory when I last asked that question HCOS/HQOS was still an MX
only thing :(

That was why the ASR920's were so handy for me.


On 12 May 2017 at 18:51, James Bensley  wrote:
> On 10 May 2017 at 22:08, Aaron Gould  wrote:
>> I also like what I've seen recently in the Juniper ACX5048 (48/72 - 10 gig
>> ports, or (6) 40 gig ports), which replaced lots of my older Cisco ME3600
>> boxes (only two 10 gig ports).
>
> Did you hit any QoS issues with this? The ACX5048 and QFX5100-48S both
> say 8 queues per port, on our MEs we have loads of queues per port, is
> this just a difference in nomenclature between vendors or really just
> 8 queues per port?
>
> Case in point, on an ME with say an Ethernet NNI port we will have an
> S-Tag per end site and multiple C-Tag per VRF/L3 VPN to that end site.
> So the ENNI port will have a simple H-QoS configuration on it. We have
> hundreds of S-Tags on an ENNI port and then on average about 3 C-Tags
> per S-Tag. Each S-Tag is shaped to the site bandwidth and then a
> policy applied to match all the C-Tags which can have to 7 traffic
> classes in it. So say 8 queues per S-tag, we have hundreds of queues
> per port.
>
> We have been looking at ACX to replace ME’s too but am I missing
> something or is the QoS capabilities of the ACX in no way comparable?
> I can't find a Juniper product that can replace the MEs at layer 2
> (48x 1G/10G ports with >10G uplinks and ME-like QoS).
>
> Cheers,
> James.
> ___
> cisco-nsp mailing list  cisco-nsp@puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/



-- 
Regards,

Mark L. Tees
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Re: [c-nsp] Cisco ASR vs Juniper

[James Bensley]

On 10 May 2017 at 22:08, Aaron Gould  wrote:
> I also like what I've seen recently in the Juniper ACX5048 (48/72 - 10 gig
> ports, or (6) 40 gig ports), which replaced lots of my older Cisco ME3600
> boxes (only two 10 gig ports).

Did you hit any QoS issues with this? The ACX5048 and QFX5100-48S both
say 8 queues per port, on our MEs we have loads of queues per port, is
this just a difference in nomenclature between vendors or really just
8 queues per port?

Case in point, on an ME with say an Ethernet NNI port we will have an
S-Tag per end site and multiple C-Tag per VRF/L3 VPN to that end site.
So the ENNI port will have a simple H-QoS configuration on it. We have
hundreds of S-Tags on an ENNI port and then on average about 3 C-Tags
per S-Tag. Each S-Tag is shaped to the site bandwidth and then a
policy applied to match all the C-Tags which can have to 7 traffic
classes in it. So say 8 queues per S-tag, we have hundreds of queues
per port.

We have been looking at ACX to replace ME’s too but am I missing
something or is the QoS capabilities of the ACX in no way comparable?
I can't find a Juniper product that can replace the MEs at layer 2
(48x 1G/10G ports with >10G uplinks and ME-like QoS).

Cheers,
James.
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Re: [c-nsp] Cisco ASR vs Juniper

[Aaron Gould]

A few things come to mind...

I enjoy my Cisco ASR9000 network running for ~5 years now... it's solid.  

I also like what I've seen recently in the Juniper ACX5048 (48/72 - 10 gig
ports, or (6) 40 gig ports), which replaced lots of my older Cisco ME3600
boxes (only two 10 gig ports).  I run MPLS L3VPN for my ISP customers.
Interestingly Juniper will automatically redistribute static routes and
connected networks into MPLS L3VPN.  Cisco requires a redistribute command.
Cisco is able to combine multiple vlan tags from same physical port into the
same bridge-domain I haven't found a way to accomplish this in Juniper
ACX5048, but I understand this is doable in Juniper MX platform.

I'm also liking what I'm seeing with my dual node CGNat boundary of Juniper
MX104's.  During testing, the MS-MIC-16G CGNat capability of Juniper seemed
nicer than the VSM-500 ASR9000 option.  There was a /27 public scope
limitation on Cisco.  Not so on Juniperyou can add public-pool /32's if
you so desire.  Also, changing public pool crashed Cisco.  Also, showing nat
translations and viewing the outside public addresses of internet hosts
wasn't nice in IOS XR you had to hunt and ask for specifics Junos
shows it easily.

In planning/discussing upgrading our existing ASR9000 ring to 100 gig, we
found that we needed to upgrade to higher CPU... I think RSP440.  But I
recall that a short life on the RSP440 meant that we were being guided to go
with the RSP880... but I think the RSP880 would cause all my trident
linecards to no longer be useable.  So we figured with that much impact me
might as well look at other vendor options too.  With that said, I'm
planning a (5) node 100 gig "super"core and have been considering both the
Cisco ASR9908 and Juniper MX960.  They both seem like solid options.

I've learned/tested a Juniper feature which is, the nicely contained logical
systems (lsys) feature of turning up pretty isolated and separate router
functions.  I understand this capability is only available on Cisco GSR/CRS
sized platforms but Juniper has it on many of its boxestesting it on
MX104 now.

- Aaron

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

[c-nsp] Cisco ASR vs Juniper

[Mark Mason]

Alright crowd...Ready the rifles and prepare for battle...Cisco ASR or Juniper. 
Cost, operability, chassis lifespan new vs. old, memory requirements, etc. So 
many details. Feel free to take the post anywhere you'd like.

Deployments:
EDGE
Pure internet router/edge with single eBGP toward ISP - 10Gb up to ISP and 10Gb 
down to Aggregation Point
MX10 w/ licensing get me 4x10Gb (but sys cap. At 20Gb?), MX40 vs ASR1001X (or 
ASR1001-HX)
+pretty vanilla installation
+not running security at this level
+just a router doing routing
+single eBGP so/so on memory requirements
+single iBGP down to Core/Agg

AGGREGATION
Internet Edge Aggregation/multiple iBGP sessions toward edge routers
MX104 or MX240 (or greater chassis lineup) vs ASR9k
+HUGE MEMORY requirements
+Multiple BGP aggregation feeds
+Installation of best routes from MULTIPLE carriers
NOTICE: This electronic mail message and any files transmitted with it are 
intended
exclusively for the individual or entity to which it is addressed. The message, 
together with any attachment, may contain confidential and/or privileged 
information.
Any unauthorized review, use, printing, saving, copying, disclosure or 
distribution 
is strictly prohibited. If you have received this message in error, please 
immediately advise the sender by reply email and delete all copies.
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/