Date
1 - 6 of 6
BSN denigrates ODP again
Ken Gray <kgray@...>
http://www.networkworld.com/community/node/83801
|
|
Luis Gomez <luis.gomez@...>
This is why we need a basic edition in ODL that should look very simple to customers.
From: tsc-bounces@... [mailto:tsc-bounces@...]
On Behalf Of Ken Gray
Sent: Thursday, September 12, 2013 10:41 AM To: tsc@... Subject: [OpenDaylight TSC] BSN denigrates ODP again
|
|
David Meyer <dmm@...>
On Fri, Sep 13, 2013 at 4:17 AM, Luis Gomez <luis.gomez@...> wrote:
This is why we need a basic edition in ODL that should look very simple toThat is no doubt true, but there is something else in the article, namely, that OF/SDN isn't a general purpose networking paradigm as it has been marketed. I've been saying this for quite some time (see for example [0] starting at slide 25) and we will want to think about what purpose of the "basic" package is. If its for R&D/NREN, that would lead to one set of functionality (likely OF/SDN). Other functionalities (e.g., network virtualization) would lead to other packaging. --dmm [0] http://www.1-4-5.net/~dmm/talks/nanog58.pdf and
|
|
Luis Gomez <luis.gomez@...>
Hi David, very good article. After reading it I think we can maybe have the following 4 packages based on SB connections and NB offered services:
toggle quoted message
Show quoted text
1) The Basic edition with only FP/SDN plugins (OF) to control forwarding devices and basic service functions - Market: R&D, FP/SDN new/greenfield operators 2) The Service edition with only FP/SDN plugins (OF) and extended service/API functions (VTN Mgr, Affinity, DDoS, OpenStack Interface) - Market: R&D, FP/SDN new/greenfield operators 3) The Transport Manager edition with FP/SDN plugins (OF) + CP/SDN plugins (BGP-LS, PCEP, I2RS, LISP, SNMP,OVSDB) to control existing transport network devices (control+forwarding) together with basic service functions + LISP Mgr - Market: Operators interested in FP/SDN that have legacy transport network devices, Operators with legacy transport network devices that just want to open their transport control (CP/SDN) 4) The Orchestrator Manager edition with FP/SDN plugins (OF) + OL/SDN plugins (REST) to control existing overlay orchestrators together with basic service functions + Dove Mgr - Applications: Operators interested in FP/SDN that have legacy overlay orchestrators, Operators with legacy overlay orchestrators that just want to open their overlay control (OL/SDN) My opinion (at least for this first release) is that we should not have many services-plugins combinations, this is why I think the Service edition should have only the OF plugin. BR/Luis -----Original Message-----
From: David Meyer [mailto:dmm@...] Sent: Thursday, September 12, 2013 3:43 PM To: Luis Gomez Cc: Ken Gray; tsc@... Subject: Re: [OpenDaylight TSC] BSN denigrates ODP again On Fri, Sep 13, 2013 at 4:17 AM, Luis Gomez <luis.gomez@...> wrote: This is why we need a basic edition in ODL that should look very simple toThat is no doubt true, but there is something else in the article, namely, that OF/SDN isn't a general purpose networking paradigm as it has been marketed. I've been saying this for quite some time (see for example [0] starting at slide 25) and we will want to think about what purpose of the "basic" package is. If its for R&D/NREN, that would lead to one set of functionality (likely OF/SDN). Other functionalities (e.g., network virtualization) would lead to other packaging. --dmm [0] http://www.1-4-5.net/~dmm/talks/nanog58.pdf and
|
|
Luis Gomez <luis.gomez@...>
Actually I would rename Transport Manager and Orchestrator Manager editions to Open Transport and Open Orchestrator editions. This sounds better to me.
toggle quoted message
Show quoted text
-----Original Message-----
From: Luis Gomez Sent: Thursday, September 12, 2013 11:09 PM To: 'David Meyer' Cc: Ken Gray; tsc@... Subject: RE: [OpenDaylight TSC] BSN denigrates ODP again Hi David, very good article. After reading it I think we can maybe have the following 4 packages based on SB connections and NB offered services: 1) The Basic edition with only FP/SDN plugins (OF) to control forwarding devices and basic service functions - Market: R&D, FP/SDN new/greenfield operators 2) The Service edition with only FP/SDN plugins (OF) and extended service/API functions (VTN Mgr, Affinity, DDoS, OpenStack Interface) - Market: R&D, FP/SDN new/greenfield operators 3) The Transport Manager edition with FP/SDN plugins (OF) + CP/SDN plugins (BGP-LS, PCEP, I2RS, LISP, SNMP,OVSDB) to control existing transport network devices (control+forwarding) together with basic service functions + LISP Mgr - Market: Operators interested in FP/SDN that have legacy transport network devices, Operators with legacy transport network devices that just want to open their transport control (CP/SDN) 4) The Orchestrator Manager edition with FP/SDN plugins (OF) + OL/SDN plugins (REST) to control existing overlay orchestrators together with basic service functions + Dove Mgr - Applications: Operators interested in FP/SDN that have legacy overlay orchestrators, Operators with legacy overlay orchestrators that just want to open their overlay control (OL/SDN) My opinion (at least for this first release) is that we should not have many services-plugins combinations, this is why I think the Service edition should have only the OF plugin. BR/Luis -----Original Message----- From: David Meyer [mailto:dmm@...] Sent: Thursday, September 12, 2013 3:43 PM To: Luis Gomez Cc: Ken Gray; tsc@... Subject: Re: [OpenDaylight TSC] BSN denigrates ODP again On Fri, Sep 13, 2013 at 4:17 AM, Luis Gomez <luis.gomez@...> wrote: This is why we need a basic edition in ODL that should look very simple toThat is no doubt true, but there is something else in the article, namely, that OF/SDN isn't a general purpose networking paradigm as it has been marketed. I've been saying this for quite some time (see for example [0] starting at slide 25) and we will want to think about what purpose of the "basic" package is. If its for R&D/NREN, that would lead to one set of functionality (likely OF/SDN). Other functionalities (e.g., network virtualization) would lead to other packaging. --dmm [0] http://www.1-4-5.net/~dmm/talks/nanog58.pdf and
|
|
David Meyer <dmm@...>
On Sep 12, 2013, at 23:44, Luis Gomez <luis.gomez@...> wrote:Agree. --dmm
|
|