Internet DRAFT - draft-zhangwu-bier-nvo3-analysis
draft-zhangwu-bier-nvo3-analysis
BIER WG Z. Zhang
Internet-Draft B. Wu
Intended status: Informational ZTE Corporation
Expires: November 25, 2017 May 24, 2017
Analysis of BIER in NVO3 network
draft-zhangwu-bier-nvo3-analysis-00
Abstract
This document analyses BIER deployment in NVO3 network. The intent
is to evaluate whether BIER could achieve some simplicity and
efficiency.
Status of This Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on November 25, 2017.
Copyright Notice
Copyright (c) 2017 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
Zhang & Wu Expires November 25, 2017 [Page 1]
Internet-Draft BIER NVO3 May 2017
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2
2. BIER in NVO3 network . . . . . . . . . . . . . . . . . . . . 2
2.1. Leaf devices as the edge BIER nodes . . . . . . . . . . . 4
2.2. NVE as edge BIER nodes . . . . . . . . . . . . . . . . . 7
3. Other Consideration . . . . . . . . . . . . . . . . . . . . . 8
4. Normative References . . . . . . . . . . . . . . . . . . . . 9
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 10
1. Introduction
As mentioned in [I-D.ietf-nvo3-mcast-framework], there is multicast
requirement in NVO3 network, such as BUM packets for infrastructure
multicast and other application-specific multicast. Besides PIM,
BIER is mentioned to be used as one of IP multicast underlay
technology in section 3, which introduces several multicast
mechanisms for NVO3 network.
Bit Index Explicit Replication (BIER) [I-D.ietf-bier-architecture] is
a new architecture for the forwarding of multicast data packets. It
provides optimal forwarding of multicast packets through a "multicast
domain". It does not require a protocol for explicitly building
multicast distribution trees, nor does it require intermediate
devices to maintain any per-flow state. When a multicast data packet
enters the BIER domain, the BIER ingress router determines the set of
BIER egress routers to which the packet needs to be sent. The BIER
ingress router then encapsulates the packet in a BIER header. The
BIER header contains a bitstring in which each bit represents exactly
one BIER egress router in the domain; to forward the packet to a
given set of egress routers, the bits corresponding to those routers
are set in the BIER header. In this way, elimination of the per-flow
state and the explicit tree-building protocols results in a
considerable simplification.
This document will give some basic analysis on how to deploy BIER in
the NVO3 network to mitigate multicast states.
2. BIER in NVO3 network
Zhang & Wu Expires November 25, 2017 [Page 2]
Internet-Draft BIER NVO3 May 2017
+--------+ +--------+
| Tenant +--+ +----| Tenant |
| System | | (') | System |
+--------+ | ................ ( ) +--------+
| +-+--+ . . +--+-+ (_)
| | NVE|--. .--| NVE| |
+--| | . . | |---+
+-+--+ . . +--+-+
/ . .
/ . L3 Overlay . +--+-++--------+
+--------+ / . Network . | NVE|| Tenant |
| Tenant +--+ . .- -| || System |
| System | . . +--+-++--------+
+--------+ ................
|
+----+
| NVE|
| |
+----+
|
|
=====================
| |
+--------+ +--------+
| Tenant | | Tenant |
| System | | System |
+--------+ +--------+
Figure 1: NVO3 Generic Reference Model
As described in [RFC8014], figure 1 is a generic reference model of
NVO3. BIER can be used in this model to provide underlay multicast
function.
Zhang & Wu Expires November 25, 2017 [Page 3]
Internet-Draft BIER NVO3 May 2017
...........................................
. ------- . -------
. ................ | Leaf| .---| NVE |
.------- . .-----| | . | |
------- .| Leaf| . . ------- . -------
| NVE |---.| |-----. . .
| | .------- . Spine . .
------- . . . .
. . . .
. . . ------- .
. . .-----| Leaf| . -------
. ................ | | .---| NVE |
. | ------- . | |
. | . -------
. ------- .
. | Leaf| .
. | | .
. ------- .
...........................................
| |
| |
------- -------
| NVE | | NVE |
| | | |
------- -------
Figure 2: Example topology of NVO3 underlay network
The architecture of L3 underlay Network can be various. As depicted
in figure 2, the Clos topology mentioned in [RFC7938] is taken as an
example to offer underlay L3 connectivity. And it is assumed that
the leaf and spine devices support BIER functionality including
exchanging BFR-id and other information and building BIER forwarding
table.
In this document, two possible ways to handle multicast packets are
identified. One of the choices is placing BIER boundaries at leaf
switches which are a general choice of implementing BIER. The other
way is putting the BIER boundary at NVE to achieve better efficiency.
2.1. Leaf devices as the edge BIER nodes
Leaf devices are taken as BIER edge node. Each leaf device is
allocated one unique BFR-id. And leaf devices are the ingress BIER
nodes and egress BIER nodes. Both the leaf and spine devices
exchange the BIER information by IGP/BGP extensions. The according
extensions are defined in [I-D.ietf-bier-ospf-bier-extensions],
[I-D.ietf-bier-isis-extensions], and [I-D.ietf-bier-idr-extensions].
Zhang & Wu Expires November 25, 2017 [Page 4]
Internet-Draft BIER NVO3 May 2017
IGMP/MLD protocol is used between NVE and leaf device like the
description in [I-D.ietf-nvo3-mcast-framework]. The BUM flows are
encapsulated corresponding multicast group address. BMLD
[I-D.pfister-bier-mld] protocol is used among all the leaf devices to
exchange multicast group information. After BMLD protocol process is
completed, each leaf devices knows the other leaf devices associated
with specific multicast group address. When one packet with
multicast group address reaches leaf device, leaf device encapsulates
the packet with BIER header which indicates all the destination leaf
devices that belong to the same multicast group.
After the BIER packet reaches the destination leaf devices through
the spine network forwarding, the destination leaf device removes the
BIER header of packet and forwards to corresponding NVEs.
So the multicast state is eliminated because of removing of multicast
tree in L3 underlay network. NVE only needs to support IGMP/MLD
protocol. But one or several multicast group addresses for a tenant
is still needed.
Zhang & Wu Expires November 25, 2017 [Page 5]
Internet-Draft BIER NVO3 May 2017
IGMP/MLD
IGMP/MLD ........................................... |
\ . ................................ . |
\ . . BIER forwarding . . |
\ . . ------- . |-------
\ . . ................ |Leaf2| .---| NVE2|
\ .------- . .-----| | . | |
------- .|Leaf1| . . ------- . -------
| NVE1|---.| |-----. . . .
| | .------- . Spine . . .
------- . . . . . .
. . . . . .
. . . . ------- .
. . . .-----|Leaf3| . -------
. . ................ | | .---| NVE3|
. . | ------- . | | |
. . | . . | -------
. . BMLD ------- . . |
. ............ |Leaf4|............ . IGMP/MLD
. | | .
. ------- .
...........................................
| IGMP/MLD |
| |
------- -------
| NVE4| | NVE5|
| | | |
------- -------
Figure 3: BIER in VNO3
For example, as illustrated in Figure 3, NVE1 needs to send BUM flows
to NVE2, NVE3, NVE4 and NVE5. The NVE supports the basic IGMP/MLD
snooping function. In most of condition, there would have to be a
separate group for each tenant, plus a separate group for each
multicast address (used for multicast applications) within a tenant.
NVE1 sends the packet to Leaf1. According the BMLD exchange Leaf1
encapsulate BIER header with the destination of Leaf2, Leaf3 and
Leaf4. After BIER forwarding the packet reaches Leaf2, Leaf3 and
Leaf4, the leaf devices remove the BIER header and send it to NVE2,
NVE3, NVE4 and NVE5.
In some use cases, there could be a big amount of leaf switch, and it
is impossible to encapsulate destination BFR-ids in one same BIER
header because of the limitation of BitStringLength, the packet
should be sent more than once to reach destination. In case the BFR-
id of Leaf2 is 28, the BFR-id of Leaf3 is 78; the BitStringLength
used in BIER encapsulation is 64, it is impossible to encapsulate the
two BFR-ids in one BIER header. In this situation, one solution is
Zhang & Wu Expires November 25, 2017 [Page 6]
Internet-Draft BIER NVO3 May 2017
Leaf1 send two copies of packet to Leaf2 and Leaf3. SI in BIER
header defined in [I-D.ietf-bier-architecture] is be used to
distinguish these two copies to deliver to the final destination.
The other solution is increase the forwarding BitStringLength to 128.
From the above analysis, BIER is quite applicable in the scenario
with limited size of leaf switches. But in a large scale of NVO3
underlay network, there is some limitation due to the
BitStringLength.
2.2. NVE as edge BIER nodes
In last section, IGMP/MLD is still needed to run between NVE and leaf
devices. And the multicast groups for Tenants and specific multicast
applications are needed.
If NVE is the edge BIER node, IGMP/MLD protocol does not need to run
between NVE and leaf device. NVE encapsulates the BIER header with
the BFR-ids of destination NVEs straightly and send it to leaf
devices. Leaf and spine devices forward the BIER packet to
destination NVEs. Destination NVEs remove the BIER header and
forwarding according to the inner encapsulation.
In this situation, leaf device does not need to be allocated BFR-id.
Every NVE should be allocated with one unique BFR-id. The BFR-id
information should be exchanged within the L3 underlay network. If
NVE supports the IGP/BGP BIER extension, NVE takes part in the BIER
information exchange. The forwarding plane will be established
easily. But in some situations NVE does not support IGP or BGP; NVE
can not take part in the BIER information exchange. So the BFR-id of
NVE should be advertised by some other method.
Besides multicast state elimination in L3 underlay network, IGMP/MLD
does not need to run between NVE and leaf devices. And BMLD does not
need to run among all the leaf devices. This function makes the BIER
deployment more simply.
Zhang & Wu Expires November 25, 2017 [Page 7]
Internet-Draft BIER NVO3 May 2017
...........................................
. ................................ .
. . BIER forwarding . .
. . ------- . -------
. . ................ |Leaf2| .---| NVE2|
.------- . .-----| | . | |
------- .|Leaf1| . . ------- . -------
| NVE1|---.| |-----. . . .
| | .------- . Spine . . .
------- . . . . . .
. . . . . .
. . . . ------- .
. . . .-----|Leaf3| . -------
. . ................ | | .---| NVE3|
. . | ------- . | |
. . | . . -------
. . ------- . .
. ............ |Leaf4|............ .
. | | .
. ------- .
...........................................
| |
| |
------- -------
| NVE4| | NVE5|
| | | |
------- -------
Figure 4: BIER in VNO3
The same example for figure 3, NVE1 should send BUM packet to NVE2,
NVE3, NVE4 and NVE5. NVE1 encapsulates BIER header with NVE2, NVE3,
NVE4 and NVE5 as destination and send it to Leaf1. Leaf1 and spine
devices forward the packet to NVE2, NVE3, NVE4 and NVE5 according to
the BIER header. The NVEs remove the BIER header and forward it.
The BitStringLength limitation also remains in this solution. And
the situation may be more seriously because of the larger number of
NVEs than leaf devices. If the BFR-id of NVE is not allocated
reasonably, in the worst situation, the forwarding efficiency is the
same with the source replication described in
[I-D.ietf-nvo3-mcast-framework] section 3.2.
3. Other Consideration
As illustrated in [RFC7938], there could be hundred thousand servers
connected by underlay network in some NVO3 network. So there are
more than thousands of NVEs and leaf devices in the network. Using
BIER as multicast underlay protocol make significant advantage
Zhang & Wu Expires November 25, 2017 [Page 8]
Internet-Draft BIER NVO3 May 2017
because of the elimination of multicast state stored in the underlay
network. But the BitStringLength limitation is one problem.
In order to achieve the optimization of BIER, the BFR-ids allocation
should be more reasonable. The BFR-id of NVE/leaf device that is
belong to one same VN could be allocated adjacent as much as
possible. So the encapsulation of BIER header can be more efficient.
Along with the number of BFR-id increasing for NVE/leaf devices,
there are thousands BIER forwarding items in the L3 underlay network.
The forwarding efficiency in the L3 underlay network should also be
considered.
4. Normative References
[I-D.ietf-bier-architecture]
Wijnands, I., Rosen, E., Dolganow, A., Przygienda, T., and
S. Aldrin, "Multicast using Bit Index Explicit
Replication", draft-ietf-bier-architecture-06 (work in
progress), April 2017.
[I-D.ietf-bier-idr-extensions]
Xu, X., Chen, M., Patel, K., Wijnands, I., and T.
Przygienda, "BGP Extensions for BIER", draft-ietf-bier-
idr-extensions-02 (work in progress), January 2017.
[I-D.ietf-bier-isis-extensions]
Ginsberg, L., Przygienda, T., Aldrin, S., and Z. Zhang,
"BIER support via ISIS", draft-ietf-bier-isis-
extensions-04 (work in progress), March 2017.
[I-D.ietf-bier-ospf-bier-extensions]
Psenak, P., Kumar, N., Wijnands, I., Dolganow, A.,
Przygienda, T., Zhang, Z., and S. Aldrin, "OSPF Extensions
for BIER", draft-ietf-bier-ospf-bier-extensions-05 (work
in progress), March 2017.
[I-D.ietf-nvo3-mcast-framework]
Ghanwani, A., Dunbar, L., McBride, M., Bannai, V., and R.
Krishnan, "A Framework for Multicast in Network
Virtualization Overlays", draft-ietf-nvo3-mcast-
framework-08 (work in progress), May 2017.
[I-D.pfister-bier-mld]
Pfister, P., Wijnands, I., Venaas, S., Wang, C., Zhang,
Z., and M. Stenberg, "BIER Ingress Multicast Flow Overlay
using Multicast Listener Discovery Protocols", draft-
pfister-bier-mld-03 (work in progress), March 2017.
Zhang & Wu Expires November 25, 2017 [Page 9]
Internet-Draft BIER NVO3 May 2017
[RFC7938] Lapukhov, P., Premji, A., and J. Mitchell, Ed., "Use of
BGP for Routing in Large-Scale Data Centers", RFC 7938,
DOI 10.17487/RFC7938, August 2016,
<http://www.rfc-editor.org/info/rfc7938>.
[RFC8014] Black, D., Hudson, J., Kreeger, L., Lasserre, M., and T.
Narten, "An Architecture for Data-Center Network
Virtualization over Layer 3 (NVO3)", RFC 8014,
DOI 10.17487/RFC8014, December 2016,
<http://www.rfc-editor.org/info/rfc8014>.
Authors' Addresses
Zheng(Sandy) Zhang
ZTE Corporation
No. 50 Software Ave, Yuhuatai Distinct
Nanjing
China
Email: zhang.zheng@zte.com.cn
Bo Wu
ZTE Corporation
No. 50 Software Ave, Yuhuatai Distinct
Nanjing
China
Email: wu.bo@zte.com.cn
Zhang & Wu Expires November 25, 2017 [Page 10]
