< draft-ietf-rtgwg-bgp-routing-large-dc-05.txt   draft-ietf-rtgwg-bgp-routing-large-dc-06.txt >
Routing Area Working Group P. Lapukhov Routing Area Working Group P. Lapukhov
Internet-Draft Facebook Internet-Draft Facebook
Intended status: Informational A. Premji Intended status: Informational A. Premji
Expires: February 1, 2016 Arista Networks Expires: February 20, 2016 Arista Networks
J. Mitchell, Ed. J. Mitchell, Ed.
July 31, 2015 August 19, 2015
Use of BGP for routing in large-scale data centers Use of BGP for routing in large-scale data centers
draft-ietf-rtgwg-bgp-routing-large-dc-05 draft-ietf-rtgwg-bgp-routing-large-dc-06
Abstract Abstract
Some network operators build and operate data centers that support Some network operators build and operate data centers that support
over one hundred thousand servers. In this document, such data over one hundred thousand servers. In this document, such data
centers are referred to as "large-scale" to differentiate them from centers are referred to as "large-scale" to differentiate them from
smaller infrastructures. Environments of this scale have a unique smaller infrastructures. Environments of this scale have a unique
set of network requirements with an emphasis on operational set of network requirements with an emphasis on operational
simplicity and network stability. This document summarizes simplicity and network stability. This document summarizes
operational experience in designing and operating large-scale data operational experience in designing and operating large-scale data
skipping to change at page 1, line 41 skipping to change at page 1, line 41
Internet-Drafts are working documents of the Internet Engineering Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet- working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/. Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress." material or to cite them other than as "work in progress."
This Internet-Draft will expire on February 1, 2016. This Internet-Draft will expire on February 20, 2016.
Copyright Notice Copyright Notice
Copyright (c) 2015 IETF Trust and the persons identified as the Copyright (c) 2015 IETF Trust and the persons identified as the
document authors. All rights reserved. document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of (http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents publication of this document. Please review these documents
skipping to change at page 2, line 34 skipping to change at page 2, line 34
3.2.1. Overview . . . . . . . . . . . . . . . . . . . . . . 7 3.2.1. Overview . . . . . . . . . . . . . . . . . . . . . . 7
3.2.2. Clos Topology Properties . . . . . . . . . . . . . . 8 3.2.2. Clos Topology Properties . . . . . . . . . . . . . . 8
3.2.3. Scaling the Clos topology . . . . . . . . . . . . . . 9 3.2.3. Scaling the Clos topology . . . . . . . . . . . . . . 9
3.2.4. Managing the Size of Clos Topology Tiers . . . . . . 10 3.2.4. Managing the Size of Clos Topology Tiers . . . . . . 10
4. Data Center Routing Overview . . . . . . . . . . . . . . . . 11 4. Data Center Routing Overview . . . . . . . . . . . . . . . . 11
4.1. Layer 2 Only Designs . . . . . . . . . . . . . . . . . . 11 4.1. Layer 2 Only Designs . . . . . . . . . . . . . . . . . . 11
4.2. Hybrid L2/L3 Designs . . . . . . . . . . . . . . . . . . 12 4.2. Hybrid L2/L3 Designs . . . . . . . . . . . . . . . . . . 12
4.3. Layer 3 Only Designs . . . . . . . . . . . . . . . . . . 12 4.3. Layer 3 Only Designs . . . . . . . . . . . . . . . . . . 12
5. Routing Protocol Selection and Design . . . . . . . . . . . . 13 5. Routing Protocol Selection and Design . . . . . . . . . . . . 13
5.1. Choosing EBGP as the Routing Protocol . . . . . . . . . . 13 5.1. Choosing EBGP as the Routing Protocol . . . . . . . . . . 13
5.2. EBGP Configuration for Clos topology . . . . . . . . . . 14 5.2. EBGP Configuration for Clos topology . . . . . . . . . . 15
5.2.1. EBGP Configuration Guidelines and Example ASN Scheme 15 5.2.1. EBGP Configuration Guidelines and Example ASN Scheme 15
5.2.2. Private Use ASNs . . . . . . . . . . . . . . . . . . 16 5.2.2. Private Use ASNs . . . . . . . . . . . . . . . . . . 16
5.2.3. Prefix Advertisement . . . . . . . . . . . . . . . . 17 5.2.3. Prefix Advertisement . . . . . . . . . . . . . . . . 17
5.2.4. External Connectivity . . . . . . . . . . . . . . . . 18 5.2.4. External Connectivity . . . . . . . . . . . . . . . . 18
5.2.5. Route Summarization at the Edge . . . . . . . . . . . 19 5.2.5. Route Summarization at the Edge . . . . . . . . . . . 19
6. ECMP Considerations . . . . . . . . . . . . . . . . . . . . . 19 6. ECMP Considerations . . . . . . . . . . . . . . . . . . . . . 19
6.1. Basic ECMP . . . . . . . . . . . . . . . . . . . . . . . 20 6.1. Basic ECMP . . . . . . . . . . . . . . . . . . . . . . . 20
6.2. BGP ECMP over Multiple ASNs . . . . . . . . . . . . . . . 21 6.2. BGP ECMP over Multiple ASNs . . . . . . . . . . . . . . . 21
6.3. Weighted ECMP . . . . . . . . . . . . . . . . . . . . . . 21 6.3. Weighted ECMP . . . . . . . . . . . . . . . . . . . . . . 21
6.4. Consistent Hashing . . . . . . . . . . . . . . . . . . . 22 6.4. Consistent Hashing . . . . . . . . . . . . . . . . . . . 22
skipping to change at page 3, line 49 skipping to change at page 3, line 49
choice and presents details of the EBGP routing design as well as choice and presents details of the EBGP routing design as well as
explores ideas for further enhancements. explores ideas for further enhancements.
This document first presents an overview of network design This document first presents an overview of network design
requirements and considerations for large-scale data centers. Then requirements and considerations for large-scale data centers. Then
traditional hierarchical data center network topologies are traditional hierarchical data center network topologies are
contrasted with Clos networks [CLOS1953] that are horizontally scaled contrasted with Clos networks [CLOS1953] that are horizontally scaled
out. This is followed by arguments for selecting EBGP with a Clos out. This is followed by arguments for selecting EBGP with a Clos
topology as the most appropriate routing protocol to meet the topology as the most appropriate routing protocol to meet the
requirements and the proposed design is described in detail. requirements and the proposed design is described in detail.
Finally, the document reviews some additional considerations and Finally, this document reviews some additional considerations and
design options. design options.
2. Network Design Requirements 2. Network Design Requirements
This section describes and summarizes network design requirements for This section describes and summarizes network design requirements for
large-scale data centers. large-scale data centers.
2.1. Bandwidth and Traffic Patterns 2.1. Bandwidth and Traffic Patterns
The primary requirement when building an interconnection network for The primary requirement when building an interconnection network for
skipping to change at page 5, line 7 skipping to change at page 5, line 7
o Driving costs down using competitive pressures, by introducing o Driving costs down using competitive pressures, by introducing
multiple network equipment vendors. multiple network equipment vendors.
In order to allow for good vendor diversity it is important to In order to allow for good vendor diversity it is important to
minimize the software feature requirements for the network elements. minimize the software feature requirements for the network elements.
This strategy provides maximum flexibility of vendor equipment This strategy provides maximum flexibility of vendor equipment
choices while enforcing interoperability using open standards. choices while enforcing interoperability using open standards.
2.3. OPEX Minimization 2.3. OPEX Minimization
Operating large-scale infrastructure could be expensive, provided Operating large-scale infrastructure can be expensive as a larger
that a larger amount of elements will statistically fail more often. amount of elements will statistically fail more often. Having a
Having a simpler design and operating using a limited software simpler design and operating using a limited software feature set
feature set minimizes software issue-related failures. minimizes software issue-related failures.
An important aspect of Operational Expenditure (OPEX) minimization is An important aspect of Operational Expenditure (OPEX) minimization is
reducing size of failure domains in the network. Ethernet networks reducing size of failure domains in the network. Ethernet networks
are known to be susceptible to broadcast or unicast traffic storms are known to be susceptible to broadcast or unicast traffic storms
that can have a dramatic impact on network performance and that can have a dramatic impact on network performance and
availability. The use of a fully routed design significantly reduces availability. The use of a fully routed design significantly reduces
the size of the data plane failure domains - i.e. limits them to the the size of the data plane failure domains - i.e. limits them to the
lowest level in the network hierarchy. However, such designs lowest level in the network hierarchy. However, such designs
introduce the problem of distributed control plane failures. This introduce the problem of distributed control plane failures. This
observation calls for simpler and less control plane protocols to observation calls for simpler and less control plane protocols to
skipping to change at page 5, line 34 skipping to change at page 5, line 34
requirements. requirements.
2.4. Traffic Engineering 2.4. Traffic Engineering
In any data center, application load balancing is a critical function In any data center, application load balancing is a critical function
performed by network devices. Traditionally, load balancers are performed by network devices. Traditionally, load balancers are
deployed as dedicated devices in the traffic forwarding path. The deployed as dedicated devices in the traffic forwarding path. The
problem arises in scaling load balancers under growing traffic problem arises in scaling load balancers under growing traffic
demand. A preferable solution would be able to scale load balancing demand. A preferable solution would be able to scale load balancing
layer horizontally, by adding more of the uniform nodes and layer horizontally, by adding more of the uniform nodes and
distributing incoming traffic across these nodes. In situation like distributing incoming traffic across these nodes. In situations like
this, an ideal choice would be to use network infrastructure itself this, an ideal choice would be to use network infrastructure itself
to distribute traffic across a group of load balancers. The to distribute traffic across a group of load balancers. The
combination of Anycast prefix advertisement [RFC4786] and Equal Cost combination of Anycast prefix advertisement [RFC4786] and Equal Cost
Multipath (ECMP) functionality can be used to accomplish this goal. Multipath (ECMP) functionality can be used to accomplish this goal.
To allow for more granular load distribution, it is beneficial for To allow for more granular load distribution, it is beneficial for
the network to support the ability to perform controlled per-hop the network to support the ability to perform controlled per-hop
traffic engineering. For example, it is beneficial to directly traffic engineering. For example, it is beneficial to directly
control the ECMP next-hop set for Anycast prefixes at every level of control the ECMP next-hop set for Anycast prefixes at every level of
network hierarchy. network hierarchy.
skipping to change at page 10, line 21 skipping to change at page 10, line 21
main reason to limit oversubscription at a single layer of the main reason to limit oversubscription at a single layer of the
network is to simplify application development that would otherwise network is to simplify application development that would otherwise
need to account for multiple bandwidth pools: within rack (Tier-3), need to account for multiple bandwidth pools: within rack (Tier-3),
between racks (Tier-2), and between clusters (Tier-1). Since between racks (Tier-2), and between clusters (Tier-1). Since
oversubscription does not have a direct relationship to the routing oversubscription does not have a direct relationship to the routing
design it is not discussed further in this document. design it is not discussed further in this document.
3.2.4. Managing the Size of Clos Topology Tiers 3.2.4. Managing the Size of Clos Topology Tiers
If a data center network size is small, it is possible to reduce the If a data center network size is small, it is possible to reduce the
number of switches in Tier-1 or Tier-2 of Clos topology by a power of number of switches in Tier-1 or Tier-2 of Clos topology by a factor
two. To understand how this could be done, take Tier-1 as an of two. To understand how this could be done, take Tier-1 as an
example. Every Tier-2 device connects to a single group of Tier-1 example. Every Tier-2 device connects to a single group of Tier-1
devices. If half of the ports on each of the Tier-1 devices are not devices. If half of the ports on each of the Tier-1 devices are not
being used then it is possible to reduce the number of Tier-1 devices being used then it is possible to reduce the number of Tier-1 devices
by half and simply map two uplinks from a Tier-2 device to the same by half and simply map two uplinks from a Tier-2 device to the same
Tier-1 device that were previously mapped to different Tier-1 Tier-1 device that were previously mapped to different Tier-1
devices. This technique maintains the same bisectional bandwidth devices. This technique maintains the same bisectional bandwidth
while reducing the number of elements in the Tier-1 layer, thus while reducing the number of elements in the Tier-1 layer, thus
saving on CAPEX. The tradeoff, in this example, is the reduction of saving on CAPEX. The tradeoff, in this example, is the reduction of
maximum DC size in terms of overall server count by half. maximum DC size in terms of overall server count by half.
skipping to change at page 11, line 23 skipping to change at page 11, line 23
Originally most data center designs used Spanning-Tree Protocol (STP) Originally most data center designs used Spanning-Tree Protocol (STP)
originally defined in [IEEE8021D-1990] for loop free topology originally defined in [IEEE8021D-1990] for loop free topology
creation, typically utilizing variants of the traditional DC topology creation, typically utilizing variants of the traditional DC topology
described in Section 3.1. At the time, many DC switches either did described in Section 3.1. At the time, many DC switches either did
not support Layer 3 routed protocols or supported it with additional not support Layer 3 routed protocols or supported it with additional
licensing fees, which played a part in the design choice. Although licensing fees, which played a part in the design choice. Although
many enhancements have been made through the introduction of Rapid many enhancements have been made through the introduction of Rapid
Spanning Tree Protocol (RSTP) in the latest revision of Spanning Tree Protocol (RSTP) in the latest revision of
[IEEE8021D-2004] and Multiple Spanning Tree Protocol (MST) specified [IEEE8021D-2004] and Multiple Spanning Tree Protocol (MST) specified
in [IEEE8021Q] that increase convergence, stability and load in [IEEE8021Q] that increase convergence, stability and load
balancing in larger topologies many of the fundamentals of the balancing in larger topologies, many of the fundamentals of the
protocol limit its applicability in large-scale DCs. STP and its protocol limit its applicability in large-scale DCs. STP and its
newer variants use an active/standby approach to path selection and newer variants use an active/standby approach to path selection and
are therefore hard to deploy in horizontally-scaled topologies as are therefore hard to deploy in horizontally-scaled topologies as
described in Section 3.2. Further, operators have had many described in Section 3.2. Further, operators have had many
experiences with large failures due to issues caused by improper experiences with large failures due to issues caused by improper
cabling, misconfiguration, or flawed software on a single device. cabling, misconfiguration, or flawed software on a single device.
These failures regularly affected the entire spanning-tree domain and These failures regularly affected the entire spanning-tree domain and
were very hard to troubleshoot due to the nature of the protocol. were very hard to troubleshoot due to the nature of the protocol.
For these reasons, and since almost all DC traffic is now IP, For these reasons, and since almost all DC traffic is now IP,
therefore requiring a Layer 3 routing protocol at the network edge therefore requiring a Layer 3 routing protocol at the network edge
for external connectivity, designs utilizing STP usually fail all of for external connectivity, designs utilizing STP usually fail all of
the requirements of large-scale DC operators. Various enhancements the requirements of large-scale DC operators. Various enhancements
to link-aggregation protocols such as [IEEE8023AD], generally known to link-aggregation protocols such as [IEEE8023AD], generally known
as Multi-Chassis Link-Aggregation (M-LAG) made it possible to use as Multi-Chassis Link-Aggregation (M-LAG) made it possible to use
Layer 2 designs with active-active network paths while relying on STP Layer 2 designs with active-active network paths while relying on STP
as the backup for loop prevention. The major downside of this as the backup for loop prevention. The major downsides of this
approach is the proprietary nature of such extensions. approach are the lack of ability to scale linearly past two in most
implementations, lack of standards based implementations, and added
failure domain risk of keeping state between the devices.
It should be noted that building large, horizontally scalable, Layer It should be noted that building large, horizontally scalable, Layer
2 only networks without STP is possible recently through the 2 only networks without STP is possible recently through the
introduction of the TRILL protocol in [RFC6325]. TRILL resolves many introduction of the TRILL protocol in [RFC6325]. TRILL resolves many
of the issues STP has for large-scale DC design however currently the of the issues STP has for large-scale DC design however due to the
maturity of the protocol, limited number of implementations, and lack of maturity of the protocol, the limited number of
requirement for new equipment that supports it has limited its implementations, and requirement for new equipment that supports it,
applicability and increased the cost of such designs. this has limited its applicability and increased the cost of such
designs.
Finally, neither TRILL nor the M-LAG approach eliminate the Finally, neither TRILL nor the M-LAG approach eliminate the
fundamental problem of the shared broadcast domain, that is so fundamental problem of the shared broadcast domain, that is so
detrimental to the operations of any Layer 2, Ethernet based detrimental to the operations of any Layer 2, Ethernet based
solutions. solutions.
4.2. Hybrid L2/L3 Designs 4.2. Hybrid L2/L3 Designs
Operators have sought to limit the impact of data plane faults and Operators have sought to limit the impact of data plane faults and
build large-scale topologies through implementing routing protocols build large-scale topologies through implementing routing protocols
in either the Tier-1 or Tier-2 parts of the network and dividing the in either the Tier-1 or Tier-2 parts of the network and dividing the
Layer-2 domain into numerous, smaller domains. This design has Layer 2 domain into numerous, smaller domains. This design has
allowed data centers to scale up, but at the cost of complexity in allowed data centers to scale up, but at the cost of complexity in
the network managing multiple protocols. For the following reasons, the network managing multiple protocols. For the following reasons,
operators have retained Layer 2 in either the access (Tier-3) or both operators have retained Layer 2 in either the access (Tier-3) or both
access and aggregation (Tier-3 and Tier-2) parts of the network: access and aggregation (Tier-3 and Tier-2) parts of the network:
o Supporting legacy applications that may require direct Layer 2 o Supporting legacy applications that may require direct Layer 2
adjacency or use non-IP protocols. adjacency or use non-IP protocols.
o Seamless mobility for virtual machines that require the o Seamless mobility for virtual machines that require the
preservation of IP addresses when a virtual machine moves to preservation of IP addresses when a virtual machine moves to
different Tier-3 switch. different Tier-3 switch.
o Simplified IP addressing = less IP subnets are required for the o Simplified IP addressing = less IP subnets are required for the
data center. data center.
o Application load balancing may require direct Layer 2 reachability o Application load balancing may require direct Layer 2 reachability
to perform certain functions such as Layer 2 Direct Server Return to perform certain functions such as Layer 2 Direct Server Return
(DSR). (DSR).
o Continued CAPEX differences between Layer-2 and Layer-3 capable o Continued CAPEX differences between Layer 2 and Layer 3 capable
switches. switches.
4.3. Layer 3 Only Designs 4.3. Layer 3 Only Designs
Network designs that leverage IP routing down to Tier-3 of the Network designs that leverage IP routing down to Tier-3 of the
network have gained popularity as well. The main benefit of these network have gained popularity as well. The main benefit of these
designs is improved network stability and scalability, as a result of designs is improved network stability and scalability, as a result of
confining L2 broadcast domains. Commonly an Interior Gateway confining L2 broadcast domains. Commonly an Interior Gateway
Protocol (IGP) such as OSPF [RFC2328] is used as the primary routing Protocol (IGP) such as OSPF [RFC2328] is used as the primary routing
protocol in such a design. As data centers grow in scale, and server protocol in such a design. As data centers grow in scale, and server
count exceeds tens of thousands, such fully routed designs have count exceeds tens of thousands, such fully routed designs have
become more attractive. become more attractive.
Choosing a Layer 3 only design greatly simplifies the network, Choosing a Layer 3 only design greatly simplifies the network,
facilitating the meeting of REQ1 and REQ2, and has widespread facilitating the meeting of REQ1 and REQ2, and has widespread
adoption in networks where large Layer 2 adjacency and larger size adoption in networks where large Layer 2 adjacency and larger size
Layer 3 subnets are not as critical compared to network scalability Layer 3 subnets are not as critical compared to network scalability
and stability. Application providers and network operators continue and stability. Application providers and network operators continue
to also develop new solutions to meet some of the requirements that to also develop new solutions to meet some of the requirements that
previously have driven large Layer 2 domains. previously have driven large Layer 2 domains by using various overlay
or tunneling techniques.
5. Routing Protocol Selection and Design 5. Routing Protocol Selection and Design
In this section the motivations for using External BGP (EBGP) as the In this section the motivations for using External BGP (EBGP) as the
single routing protocol for data center networks having a Layer 3 single routing protocol for data center networks having a Layer 3
protocol design and Clos topology are reviewed. Then, a practical protocol design and Clos topology are reviewed. Then, a practical
approach for designing an EBGP based network is provided. approach for designing an EBGP based network is provided.
5.1. Choosing EBGP as the Routing Protocol 5.1. Choosing EBGP as the Routing Protocol
skipping to change at page 14, line 12 skipping to change at page 14, line 14
flow-control, BGP simply relies on TCP as the underlying flow-control, BGP simply relies on TCP as the underlying
transport. This fulfills REQ2 and REQ3. transport. This fulfills REQ2 and REQ3.
o BGP information flooding overhead is less when compared to link- o BGP information flooding overhead is less when compared to link-
state IGPs. Since every BGP router calculates and propagates only state IGPs. Since every BGP router calculates and propagates only
the best-path selected, a network failure is masked as soon as the the best-path selected, a network failure is masked as soon as the
BGP speaker finds an alternate path, which exists when highly BGP speaker finds an alternate path, which exists when highly
symmetric topologies, such as Clos, are coupled with EBGP only symmetric topologies, such as Clos, are coupled with EBGP only
design. In contrast, the event propagation scope of a link-state design. In contrast, the event propagation scope of a link-state
IGP is an entire area, regardless of the failure type. This meets IGP is an entire area, regardless of the failure type. This meets
REQ3 and REQ4. It is worth mentioning that all widely deployed REQ3 and REQ4. It is also worth mentioning that all widely
link-state IGPs also feature periodic refreshes of routing deployed link-state IGPs feature periodic refreshes of routing
information, while BGP does not expire routing state, even if this information, even if this rarely causes impact to modern router
rarely causes significant impact to modern router control planes. control planes, while BGP does not expire routing state.
o BGP supports third-party (recursively resolved) next-hops. This o BGP supports third-party (recursively resolved) next-hops. This
allows for manipulating multipath to be non-ECMP based or allows for manipulating multipath to be non-ECMP based or
forwarding based on application-defined paths, through forwarding based on application-defined paths, through
establishment of a peering session with an application establishment of a peering session with an application
"controller" which can inject routing information into the system, "controller" which can inject routing information into the system,
satisfying REQ5. OSPF provides similar functionality using satisfying REQ5. OSPF provides similar functionality using
concepts such as "Forwarding Address", but with more difficulty in concepts such as "Forwarding Address", but with more difficulty in
implementation and far less control of information propagation implementation and far less control of information propagation
scope. scope.
skipping to change at page 14, line 44 skipping to change at page 14, line 46
Using a traditional single flooding domain, which most DC designs Using a traditional single flooding domain, which most DC designs
utilize, under certain failure conditions may pick up unwanted utilize, under certain failure conditions may pick up unwanted
lengthy paths, e.g. traversing multiple Tier-2 devices. lengthy paths, e.g. traversing multiple Tier-2 devices.
o EBGP configuration that is implemented with minimal routing policy o EBGP configuration that is implemented with minimal routing policy
is easier to troubleshoot for network reachability issues. In is easier to troubleshoot for network reachability issues. In
most implementations, it is straightforward to view contents of most implementations, it is straightforward to view contents of
BGP Loc-RIB and compare it to the router's RIB. Also, in most BGP Loc-RIB and compare it to the router's RIB. Also, in most
implementations an operator can view every BGP neighbors Adj-RIB- implementations an operator can view every BGP neighbors Adj-RIB-
In and Adj-RIB-Out structures and therefore incoming and outgoing In and Adj-RIB-Out structures and therefore incoming and outgoing
NRLI information can be easily correlated on both sides of a BGP NLRI information can be easily correlated on both sides of a BGP
session. Thus, BGP satisfies REQ3. session. Thus, BGP satisfies REQ3.
5.2. EBGP Configuration for Clos topology 5.2. EBGP Configuration for Clos topology
Clos topologies that have more than 5 stages are very uncommon due to Clos topologies that have more than 5 stages are very uncommon due to
the large numbers of interconnects required by such a design. the large numbers of interconnects required by such a design.
Therefore, the examples below are made with reference to the 5-stage Therefore, the examples below are made with reference to the 5-stage
Clos topology (in unfolded state). Clos topology (in unfolded state).
5.2.1. EBGP Configuration Guidelines and Example ASN Scheme 5.2.1. EBGP Configuration Guidelines and Example ASN Scheme
skipping to change at page 15, line 21 skipping to change at page 15, line 28
point links interconnecting the network nodes, no multi-hop or point links interconnecting the network nodes, no multi-hop or
loopback sessions are used even in the case of multiple links loopback sessions are used even in the case of multiple links
between the same pair of nodes. between the same pair of nodes.
o Private Use ASNs from the range 64512-65534 are used so as to o Private Use ASNs from the range 64512-65534 are used so as to
avoid ASN conflicts. avoid ASN conflicts.
o A single ASN is allocated to all of the Clos topology's Tier-1 o A single ASN is allocated to all of the Clos topology's Tier-1
devices. devices.
o A unique ASN is allocated per each group of Tier-2 devices. o A unique ASN is allocated to each set of Tier-2 devices in the
same cluster.
o A unique ASN is allocated to every Tier-3 device (e.g. ToR) in o A unique ASN is allocated to every Tier-3 device (e.g. ToR) in
this topology. this topology.
ASN 65534 ASN 65534
+---------+ +---------+
| +-----+ | | +-----+ |
| | | | | | | |
+-|-| |-|-+ +-|-| |-|-+
| | +-----+ | | | | +-----+ | |
skipping to change at page 18, line 5 skipping to change at page 18, line 5
subnets in a Clos topology results in route black-holing under a subnets in a Clos topology results in route black-holing under a
single link failure (e.g. between Tier-2 and Tier-3 devices) and single link failure (e.g. between Tier-2 and Tier-3 devices) and
hence must be avoided. The use of peer links within the same tier to hence must be avoided. The use of peer links within the same tier to
resolve the black-holing problem by providing "bypass paths" is resolve the black-holing problem by providing "bypass paths" is
undesirable due to O(N^2) complexity of the peering mesh and waste of undesirable due to O(N^2) complexity of the peering mesh and waste of
ports on the devices. An alternative to the full-mesh of peer-links ports on the devices. An alternative to the full-mesh of peer-links
would be using a simpler bypass topology, e.g. a "ring" as described would be using a simpler bypass topology, e.g. a "ring" as described
in [FB4POST], but such a topology adds extra hops and has very in [FB4POST], but such a topology adds extra hops and has very
limited bisection bandwidth, in addition requiring special tweaks to limited bisection bandwidth, in addition requiring special tweaks to
make BGP routing work - such as possibly splitting every device into make BGP routing work - such as possibly splitting every device into
an ASN on its own. The section Section 8.2 introduces another, less an ASN on its own. Later in this document, Section 8.2 introduces a
intrusive, method for performing a limited form route summarization less intrusive method for performing a limited form route
in Clos networks and the discusses the associated trade-offs. summarization in Clos networks and discusses it's associated trade-
offs.
5.2.4. External Connectivity 5.2.4. External Connectivity
A dedicated cluster (or clusters) in the Clos topology could be used A dedicated cluster (or clusters) in the Clos topology could be used
for the purpose of connecting to the Wide Area Network (WAN) edge for the purpose of connecting to the Wide Area Network (WAN) edge
devices, or WAN Routers. Tier-3 devices in such cluster would be devices, or WAN Routers. Tier-3 devices in such cluster would be
replaced with WAN routers, and EBGP peering would be used again, replaced with WAN routers, and EBGP peering would be used again,
though WAN routers are likely to belong to a public ASN if Internet though WAN routers are likely to belong to a public ASN if Internet
connectivity is required in the design. The Tier-2 devices in such a connectivity is required in the design. The Tier-2 devices in such a
dedicated cluster will be referred to as "Border Routers" in this dedicated cluster will be referred to as "Border Routers" in this
skipping to change at page 19, line 24 skipping to change at page 19, line 24
due to the lack of peer links inside every tier. due to the lack of peer links inside every tier.
However, it is possible to lift this restriction for the Border However, it is possible to lift this restriction for the Border
Routers, by devising a different connectivity model for these Routers, by devising a different connectivity model for these
devices. There are two options possible: devices. There are two options possible:
o Interconnect the Border Routers using a full-mesh of physical o Interconnect the Border Routers using a full-mesh of physical
links or using any other "peer-mesh" topology, such as ring or links or using any other "peer-mesh" topology, such as ring or
hub-and-spoke. Configure BGP accordingly on all Border Leafs to hub-and-spoke. Configure BGP accordingly on all Border Leafs to
exchange network reachability information - e.g. by adding a mesh exchange network reachability information - e.g. by adding a mesh
of iBGP sessions. The interconnecting peer links need to be of IBGP sessions. The interconnecting peer links need to be
appropriately sized for traffic that will be present in the case appropriately sized for traffic that will be present in the case
of a device or link failure underneath the Border Routers. of a device or link failure underneath the Border Routers.
o Tier-1 devices may have additional physical links provisioned o Tier-1 devices may have additional physical links provisioned
toward the Border Routers (which are Tier-2 devices from the toward the Border Routers (which are Tier-2 devices from the
perspective of Tier-1). Specifically, if protection from a single perspective of Tier-1). Specifically, if protection from a single
link or node failure is desired, each Tier-1 devices would have to link or node failure is desired, each Tier-1 devices would have to
connect to at least two Border Routers. This puts additional connect to at least two Border Routers. This puts additional
requirements on the port count for Tier-1 devices and Border requirements on the port count for Tier-1 devices and Border
Routers, potentially making it a non-uniform, larger port count, Routers, potentially making it a non-uniform, larger port count,
device with the other devices in the Clos. This also reduces the device compared with the other devices in the Clos. This also
number of ports available to "regular" Tier-2 switches and hence reduces the number of ports available to "regular" Tier-2 switches
the number of clusters that could be interconnected via Tier-1 and hence the number of clusters that could be interconnected via
layer. Tier-1 layer.
If any of the above options are implemented, it is possible to If any of the above options are implemented, it is possible to
perform route summarization at the Border Routers toward the WAN perform route summarization at the Border Routers toward the WAN
network core without risking a routing black-hole condition under a network core without risking a routing black-hole condition under a
single link failure. Both of the options would result in non-uniform single link failure. Both of the options would result in non-uniform
topology as additional links have to be provisioned on some network topology as additional links have to be provisioned on some network
devices. devices.
6. ECMP Considerations 6. ECMP Considerations
skipping to change at page 21, line 10 skipping to change at page 21, line 10
able to connect to multitude of Tier-1 devices if route summarization able to connect to multitude of Tier-1 devices if route summarization
at Border Router level is implemented as described in Section 5.2.5. at Border Router level is implemented as described in Section 5.2.5.
If a device's hardware does not support wider ECMP, logical link- If a device's hardware does not support wider ECMP, logical link-
grouping (link-aggregation at layer 2) could be used to provide grouping (link-aggregation at layer 2) could be used to provide
"hierarchical" ECMP (Layer 3 ECMP followed by Layer 2 ECMP) to "hierarchical" ECMP (Layer 3 ECMP followed by Layer 2 ECMP) to
compensate for fan-out limitations. Such approach, however, compensate for fan-out limitations. Such approach, however,
increases the risk of flow polarization, as less entropy will be increases the risk of flow polarization, as less entropy will be
available to the second stage of ECMP. available to the second stage of ECMP.
Most BGP implementations declare paths to be equal from ECMP Most BGP implementations declare paths to be equal from ECMP
perspective if they match up to and including step (e) perspective if they match up to and including step (e) in
Section 9.1.2.2 of [RFC4271]. In the proposed network design there Section 9.1.2.2 of [RFC4271]. In the proposed network design there
is no underlying IGP, so all IGP costs are assumed to be zero or is no underlying IGP, so all IGP costs are assumed to be zero or
otherwise the same value across all paths and policies may be applied otherwise the same value across all paths and policies may be applied
as necessary to equalize BGP attributes that vary in vendor defaults, as necessary to equalize BGP attributes that vary in vendor defaults,
such as MED and origin code. For historical reasons it is also such as MED and origin code. For historical reasons it is also
useful to not use 0 as the equalized MED value; this and some other useful to not use 0 as the equalized MED value; this and some other
useful BGP information is available in [RFC4277] . Routing loops are useful BGP information is available in [RFC4277] . Routing loops are
unlikely due to the BGP best-path selection process which prefers unlikely due to the BGP best-path selection process which prefers
shorter AS_PATH length, and longer paths through the Tier-1 devices shorter AS_PATH length, and longer paths through the Tier-1 devices
which don't allow their own ASN in the path and have the same ASN are which don't allow their own ASN in the path and have the same ASN are
skipping to change at page 22, line 7 skipping to change at page 22, line 7
and send more traffic over paths that have more capacity. The and send more traffic over paths that have more capacity. The
prefixes that require weighted ECMP would have to be injected using prefixes that require weighted ECMP would have to be injected using
remote BGP speaker (central agent) over a multihop session as remote BGP speaker (central agent) over a multihop session as
described further in Section 8.1. If support in implementations is described further in Section 8.1. If support in implementations is
available, weight-distribution for multiple BGP paths could be available, weight-distribution for multiple BGP paths could be
signaled using the technique described in signaled using the technique described in
[I-D.ietf-idr-link-bandwidth]. [I-D.ietf-idr-link-bandwidth].
6.4. Consistent Hashing 6.4. Consistent Hashing
It is often desirable to have the hashing function used to ECMP to be It is often desirable to have the hashing function used for ECMP to
consistent (see [CONS-HASH]), to minimizing the impact on flow to be consistent (see [CONS-HASH]), to minimize the impact on flow to
next-hop affinity changes when a next-hop is added or removed to ECMP next-hop affinity changes when a next-hop is added or removed to ECMP
group. This could be used if the network device is used as a load group. This could be used if the network device is used as a load
balancer, mapping flows toward multiple destinations - in this case, balancer, mapping flows toward multiple destinations - in this case,
losing or adding a destination will not have detrimental effect of losing or adding a destination will not have detrimental effect of
currently established flows. One particular recommendation on currently established flows. One particular recommendation on
implementing consistent hashing is provided in [RFC2992], though implementing consistent hashing is provided in [RFC2992], though
other implementations are possible. This functionality could be other implementations are possible. This functionality could be
naturally combined with weighted ECMP, with the impact of the next- naturally combined with weighted ECMP, with the impact of the next-
hop changes being proportional to the weight of the given next-hop. hop changes being proportional to the weight of the given next-hop.
The downside of consistent hashing is increased load on hardware The downside of consistent hashing is increased load on hardware
skipping to change at page 22, line 50 skipping to change at page 22, line 50
convergence delays, in the order of multiple seconds (on many BGP convergence delays, in the order of multiple seconds (on many BGP
implementations the minimum configurable BGP hold timer value is implementations the minimum configurable BGP hold timer value is
three seconds). However, many BGP implementations can shut down three seconds). However, many BGP implementations can shut down
local EBGP peering sessions in response to the "link down" event for local EBGP peering sessions in response to the "link down" event for
the outgoing interface used for BGP peering. This feature is the outgoing interface used for BGP peering. This feature is
sometimes called as "fast fallover". Since links in modern data sometimes called as "fast fallover". Since links in modern data
centers are predominantly point-to-point fiber connections, a centers are predominantly point-to-point fiber connections, a
physical interface failure is often detected in milliseconds and physical interface failure is often detected in milliseconds and
subsequently triggers a BGP re-convergence. subsequently triggers a BGP re-convergence.
Ethernet technologies may support failure signaling or detection Ethernet links may support failure signaling or detection standards
standards such as Connectivity Fault Management (CFM) as described in such as Connectivity Fault Management (CFM) as described in
[IEEE8021Q], which may make failure detection more robust. [IEEE8021Q], which may make failure detection more robust.
Alternatively, some platforms may support Bidirectional Forwarding Alternatively, some platforms may support Bidirectional Forwarding
Detection (BFD) [RFC5880] to allow for sub-second failure detection Detection (BFD) [RFC5880] to allow for sub-second failure detection
and fault signaling to the BGP process. However, use of either of and fault signaling to the BGP process. However, use of either of
these presents additional requirements to vendor software and these presents additional requirements to vendor software and
possibly hardware, and may contradict REQ1. Until recently with possibly hardware, and may contradict REQ1. Until recently with
[RFC7130], BFD also did not allow detection of a single member link [RFC7130], BFD also did not allow detection of a single member link
failure on a LAG, which would have limited it's usefulness in some failure on a LAG, which would have limited its usefulness in some
designs. designs.
7.2. Event Propagation Timing 7.2. Event Propagation Timing
In the proposed design the impact of BGP Minimum Route Advertisement In the proposed design the impact of BGP Minimum Route Advertisement
Interval (MRAI) timer (See section 9.2.1.1 of [RFC4271]) should be Interval (MRAI) timer (See section 9.2.1.1 of [RFC4271]) should be
considered. Per the standard it is required for BGP implementations considered. Per the standard it is required for BGP implementations
to space out consecutive BGP UPDATE messages by at least MRAI to space out consecutive BGP UPDATE messages by at least MRAI
seconds, which is often a configurable value. The initial BGP UPDATE seconds, which is often a configurable value. The initial BGP UPDATE
messages after an event carrying withdrawn routes are commonly not messages after an event carrying withdrawn routes are commonly not
skipping to change at page 25, line 23 skipping to change at page 25, line 23
ECMP groups for all IP prefixes from non-local cluster. The ECMP groups for all IP prefixes from non-local cluster. The
Tier-3 devices are once again not involved in the re-convergence Tier-3 devices are once again not involved in the re-convergence
process, but may receive "implicit withdraws" as described above. process, but may receive "implicit withdraws" as described above.
Even though in case of such failures multiple IP prefixes will have Even though in case of such failures multiple IP prefixes will have
to be reprogrammed in the FIB, it is worth noting that ALL of these to be reprogrammed in the FIB, it is worth noting that ALL of these
prefixes share a single ECMP group on Tier-2 device. Therefore, in prefixes share a single ECMP group on Tier-2 device. Therefore, in
the case of implementations with a hierarchical FIB, only a single the case of implementations with a hierarchical FIB, only a single
change has to be made to the FIB. Hierarchical FIB here means FIB change has to be made to the FIB. Hierarchical FIB here means FIB
structure where the next-hop forwarding information is stored structure where the next-hop forwarding information is stored
separately from the prefix lookup table, and the latter only store separately from the prefix lookup table, and the latter only stores
pointers to the respective forwarding information. pointers to the respective forwarding information.
Even though BGP offers reduced failure scope for some cases, further Even though BGP offers reduced failure scope for some cases, further
reduction of the fault domain using summarization is not always reduction of the fault domain using summarization is not always
possible with the proposed design, since using this technique may possible with the proposed design, since using this technique may
create routing black-holes as mentioned previously. Therefore, the create routing black-holes as mentioned previously. Therefore, the
worst control plane failure impact scope is the network as a whole, worst control plane failure impact scope is the network as a whole,
for instance in a case of a link failure between Tier-2 and Tier-3 for instance in a case of a link failure between Tier-2 and Tier-3
devices. The amount of impacted prefixes in this case would be much devices. The amount of impacted prefixes in this case would be much
less than in the case of a failure in the upper layers of a Clos less than in the case of a failure in the upper layers of a Clos
skipping to change at page 26, line 11 skipping to change at page 26, line 11
Tier-2 will bounce it back again using the default route. This Tier-2 will bounce it back again using the default route. This
micro-loop will last for the duration of time it takes the upstream micro-loop will last for the duration of time it takes the upstream
device to fully update its forwarding tables. device to fully update its forwarding tables.
To minimize impact of the micro-loops, Tier-2 and Tier-1 switches can To minimize impact of the micro-loops, Tier-2 and Tier-1 switches can
be configured with static "discard" or "null" routes that will be be configured with static "discard" or "null" routes that will be
more specific than the default route for prefixes missing during more specific than the default route for prefixes missing during
network convergence. For Tier-2 switches, the discard route should network convergence. For Tier-2 switches, the discard route should
be a summary route, covering all server subnets of the underlying be a summary route, covering all server subnets of the underlying
Tier-3 devices. For Tier-1 devices, the discard route should be a Tier-3 devices. For Tier-1 devices, the discard route should be a
summary covering the server IP address subnet allocated for the whole summary covering the server IP address subnets allocated for the
data center. Those discard routes will only take precedence for the whole data center. Those discard routes will only take precedence
duration of network convergence, until the device learns a more for the duration of network convergence, until the device learns a
specific prefix via a new path. more specific prefix via a new path.
8. Additional Options for Design 8. Additional Options for Design
8.1. Third-party Route Injection 8.1. Third-party Route Injection
BGP allows for a "third-party", i.e. directly attached, BGP speaker BGP allows for a "third-party", i.e. directly attached, BGP speaker
to inject routes anywhere in the network topology, meeting REQ5. to inject routes anywhere in the network topology, meeting REQ5.
This can be achieved by peering via a multihop BGP session with some This can be achieved by peering via a multihop BGP session with some
or even all devices in the topology. Furthermore, BGP diverse path or even all devices in the topology. Furthermore, BGP diverse path
distribution [RFC6774] could be used to inject multiple BGP next hops distribution [RFC6774] could be used to inject multiple BGP next hops
for the same prefix to facilitate load balancing, or using the BGP for the same prefix to facilitate load balancing, or using the BGP
ADD-PATH capability [I-D.ietf-idr-add-paths] if supported by the ADD-PATH capability [I-D.ietf-idr-add-paths] if supported by the
implementation. Unfortunately, in many implementations ADD-PATH has implementation. Unfortunately, in many implementations ADD-PATH has
been found to only support IBGP properly due to the use cases it was been found to only support IBGP properly due to the use cases it was
originally optimized for, which limits the "third-party" peering to originally optimized for, which limits the "third-party" peering to
iBGP only, if the feature is used. IBGP only, if the feature is used.
To implement route injection in the proposed design, a third-party To implement route injection in the proposed design, a third-party
BGP speaker may peer with Tier-3 and Tier-1 switches, injecting the BGP speaker may peer with Tier-3 and Tier-1 switches, injecting the
same prefix, but using a special set of BGP next-hops for Tier-1 same prefix, but using a special set of BGP next-hops for Tier-1
devices. Those next-hops are assumed to resolve recursively via BGP, devices. Those next-hops are assumed to resolve recursively via BGP,
and could be, for example, IP addresses on Tier-3 devices. The and could be, for example, IP addresses on Tier-3 devices. The
resulting forwarding table programming could provide desired traffic resulting forwarding table programming could provide desired traffic
proportion distribution among different clusters. proportion distribution among different clusters.
8.2. Route Summarization within Clos Topology 8.2. Route Summarization within Clos Topology
skipping to change at page 27, line 8 skipping to change at page 27, line 8
devices. However, some operators may find route aggregation devices. However, some operators may find route aggregation
desirable to improve control plane stability. desirable to improve control plane stability.
If planning on using any technique to summarize within the topology If planning on using any technique to summarize within the topology
modeling of the routing behavior and potential for black-holing modeling of the routing behavior and potential for black-holing
should be done not only for single or multiple link failures, but should be done not only for single or multiple link failures, but
also fiber pathway failures or optical domain failures if the also fiber pathway failures or optical domain failures if the
topology extends beyond a physical location. Simple modeling can be topology extends beyond a physical location. Simple modeling can be
done by checking the reachability on devices doing summarization done by checking the reachability on devices doing summarization
under the condition of a link or pathway failure between a set of under the condition of a link or pathway failure between a set of
devices in every Tier as well as to the WAN routers if external devices in every tier as well as to the WAN routers if external
connectivity is present. connectivity is present.
Route summarization would be possible with a small modification to Route summarization would be possible with a small modification to
the network topology, though the trade-off would be reduction of the the network topology, though the trade-off would be reduction of the
total size of the network as well as network congestion under total size of the network as well as network congestion under
specific failures. This approach is very similar to the technique specific failures. This approach is very similar to the technique
described above, which allows Border Routers to summarize the entire described above, which allows Border Routers to summarize the entire
data center address space. data center address space.
8.2.1. Collapsing Tier-1 Devices Layer 8.2.1. Collapsing Tier-1 Devices Layer
skipping to change at page 28, line 35 skipping to change at page 28, line 35
8.2.2. Simple Virtual Aggregation 8.2.2. Simple Virtual Aggregation
A completely different approach to route summarization is possible, A completely different approach to route summarization is possible,
provided that the main goal is to reduce the FIB pressure, while provided that the main goal is to reduce the FIB pressure, while
allowing the control plane to disseminate full routing information. allowing the control plane to disseminate full routing information.
Firstly, it could be easily noted that in many cases multiple Firstly, it could be easily noted that in many cases multiple
prefixes, some of which are less specific, share the same set of the prefixes, some of which are less specific, share the same set of the
next-hops (same ECMP group). For example, looking from the next-hops (same ECMP group). For example, looking from the
perspective of a Tier-3 devices, all routes learned from upstream perspective of a Tier-3 devices, all routes learned from upstream
Tier-2's, including the default route, will share the same set of BGP Tier-2's, including the default route, will share the same set of BGP
next-hops, provided that there is no failures in the network. This next-hops, provided that there are no failures in the network. This
makes it possible to use the technique similar to described in makes it possible to use the technique similar to described in
[RFC6769] and only install the least specific route in the FIB, [RFC6769] and only install the least specific route in the FIB,
ignoring more specific routes if they share the same next-hop set. ignoring more specific routes if they share the same next-hop set.
For example, under normal network conditions, only the default route For example, under normal network conditions, only the default route
need to be programmed into FIB. need to be programmed into FIB.
Furthermore, if the Tier-2 devices are configured with summary Furthermore, if the Tier-2 devices are configured with summary
prefixes covering all of their attached Tier-3 device's prefixes the prefixes covering all of their attached Tier-3 device's prefixes the
same logic could be applied in Tier-1 devices as well, and, by same logic could be applied in Tier-1 devices as well, and, by
induction to Tier-2/Tier-3 switches in different clusters. These induction to Tier-2/Tier-3 switches in different clusters. These
skipping to change at page 30, line 11 skipping to change at page 30, line 11
10. IANA Considerations 10. IANA Considerations
This document includes no request to IANA. This document includes no request to IANA.
11. Acknowledgements 11. Acknowledgements
This publication summarizes work of many people who participated in This publication summarizes work of many people who participated in
developing, testing and deploying the proposed network design, some developing, testing and deploying the proposed network design, some
of whom were George Chen, Parantap Lahiri, Dave Maltz, Edet Nkposong, of whom were George Chen, Parantap Lahiri, Dave Maltz, Edet Nkposong,
Robert Toomey, and Lihua Yuan. Authors would also like to thank Robert Toomey, and Lihua Yuan. Authors would also like to thank
Linda Dunbar, Susan Hares, Danny McPherson, Robert Raszuk and Russ Linda Dunbar, Anoop Ghanwani, Susan Hares, Danny McPherson, Robert
White for reviewing the document and providing valuable feedback and Raszuk and Russ White for reviewing this document and providing
Mary Mitchell for grammar and style suggestions. valuable feedback and Mary Mitchell for initial grammar and style
suggestions.
12. References 12. References
12.1. Normative References 12.1. Normative References
[RFC4271] Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A [RFC4271] Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A
Border Gateway Protocol 4 (BGP-4)", RFC 4271, Border Gateway Protocol 4 (BGP-4)", RFC 4271,
DOI 10.17487/RFC4271, January 2006, DOI 10.17487/RFC4271, January 2006,
<http://www.rfc-editor.org/info/rfc4271>. <http://www.rfc-editor.org/info/rfc4271>.
skipping to change at page 32, line 14 skipping to change at page 32, line 14
[I-D.ietf-idr-link-bandwidth] [I-D.ietf-idr-link-bandwidth]
Mohapatra, P. and R. Fernando, "BGP Link Bandwidth Mohapatra, P. and R. Fernando, "BGP Link Bandwidth
Extended Community", draft-ietf-idr-link-bandwidth-06 Extended Community", draft-ietf-idr-link-bandwidth-06
(work in progress), January 2013. (work in progress), January 2013.
[CLOS1953] [CLOS1953]
Clos, C., "A Study of Non-Blocking Switching Networks: Clos, C., "A Study of Non-Blocking Switching Networks:
Bell System Technical Journal Vol. 32(2)", March 1953. Bell System Technical Journal Vol. 32(2)", March 1953.
[HADOOP] Apache, , "Apache HaDoop", July 2015, [HADOOP] Apache, , "Apache HaDoop", August 2015,
<https://hadoop.apache.org/>. <https://hadoop.apache.org/>.
[GREENBERG2009] [GREENBERG2009]
Greenberg, A., Hamilton, J., and D. Maltz, "The Cost of a Greenberg, A., Hamilton, J., and D. Maltz, "The Cost of a
Cloud: Research Problems in Data Center Networks", January Cloud: Research Problems in Data Center Networks", January
2009. 2009.
[IEEE8021D-1990] [IEEE8021D-1990]
IEEE 802.1D, , "IEEE Standard for Local and Metropolitan IEEE 802.1D, , "IEEE Standard for Local and Metropolitan
Area Networks--Media access control (MAC) Bridges", May Area Networks--Media access control (MAC) Bridges", May
skipping to change at page 32, line 46 skipping to change at page 32, line 46
[INTERCON] [INTERCON]
Dally, W. and B. Towles, "Principles and Practices of Dally, W. and B. Towles, "Principles and Practices of
Interconnection Networks", ISBN 978-0122007514, January Interconnection Networks", ISBN 978-0122007514, January
2004. 2004.
[ALFARES2008] [ALFARES2008]
Al-Fares, M., Loukissas, A., and A. Vahdat, "A Scalable, Al-Fares, M., Loukissas, A., and A. Vahdat, "A Scalable,
Commodity Data Center Network Architecture", August 2008. Commodity Data Center Network Architecture", August 2008.
[IANA.AS] IANA, , "Autonomous System (AS) Numbers", July 2015, [IANA.AS] IANA, , "Autonomous System (AS) Numbers", August 2015,
<http://www.iana.org/assignments/as-numbers/>. <http://www.iana.org/assignments/as-numbers/>.
[IEEE8023AD] [IEEE8023AD]
IEEE 802.3ad, , "IEEE Standard for Link aggregation for IEEE 802.3ad, , "IEEE Standard for Link aggregation for
parallel links", October 2000. parallel links", October 2000.
[ALLOWASIN] [ALLOWASIN]
Cisco Systems, , "Allowas-in Feature in BGP Configuration Cisco Systems, , "Allowas-in Feature in BGP Configuration
Example", February 2015, Example", February 2015,
<http://www.cisco.com/c/en/us/support/docs/ip/border- <http://www.cisco.com/c/en/us/support/docs/ip/border-
 End of changes. 33 change blocks. 
56 lines changed or deleted 63 lines changed or added

This html diff was produced by rfcdiff 1.48. The latest version is available from http://tools.ietf.org/tools/rfcdiff/