RFC 7654

Internet Engineering Task Force (IETF)                          S. Banks
Request for Comments: 7654                                VSS Monitoring
Category: Informational                                      F. Calabria
ISSN: 2070-1721                                            Cisco Systems
                                                              G. Czirjak
                                                               R. Machat
                                                        Juniper Networks
                                                            October 2015

    Benchmarking Methodology for In-Service Software Upgrade (ISSU)


   Modern forwarding devices attempt to minimize any control- and data-
   plane disruptions while performing planned software changes by
   implementing a technique commonly known as In-Service Software
   Upgrade (ISSU).  This document specifies a set of common
   methodologies and procedures designed to characterize the overall
   behavior of a Device Under Test (DUT), subject to an ISSU event.

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Not all documents
   approved by the IESG are a candidate for any level of Internet
   Standard; see Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at

Banks, et al.                 Informational                     [Page 1]

RFC 7654              Benchmarking Software Upgrade         October 2015

Copyright Notice

   Copyright (c) 2015 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.

Table of Contents

   1. Introduction ....................................................3
   2. Conventions Used in This Document ...............................4
   3. Generic ISSU Process, Phased Approach ...........................4
      3.1. Software Download ..........................................5
      3.2. Software Staging ...........................................6
      3.3. Upgrade Run ................................................6
      3.4. Upgrade Acceptance .........................................7
   4. Test Methodology ................................................7
      4.1. Test Topology ..............................................7
      4.2. Load Model .................................................8
   5. ISSU Test Methodology ...........................................9
      5.1. Pre-ISSU Recommended Verifications .........................9
      5.2. Software Staging ...........................................9
      5.3. Upgrade Run ...............................................10
      5.4. Post-ISSU Verification ....................................11
      5.5. ISSU under Negative Stimuli ...............................12
   6. ISSU Abort and Rollback ........................................12
   7. Final Report: Data Presentation and Analysis ...................13
      7.1. Data Collection Considerations ............................14
   8. Security Considerations ........................................15
   9. References .....................................................15
      9.1. Normative References ......................................15
      9.2. Informative References ....................................16
   Acknowledgments ...................................................16
   Authors' Addresses ................................................16

Banks, et al.                 Informational                     [Page 2]

RFC 7654              Benchmarking Software Upgrade         October 2015

1.  Introduction

   As required by most Service Provider (SP) network operators, ISSU
   functionality has been implemented by modern forwarding devices to
   upgrade or downgrade from one software version to another with a goal
   of eliminating the downtime of the router and/or the outage of
   service.  However, it is noted that while most operators desire
   complete elimination of downtime, minimization of downtime and
   service degradation is often the expectation.

   The ISSU operation may apply in terms of an atomic version change of
   the entire system software or it may be applied in a more modular
   sense, such as for a patch or maintenance upgrade.  The procedure
   described herein may be used to verify either approach, as may be
   supported by the vendor hardware and software.

   In support of this document, the desired behavior for an ISSU
   operation can be summarized as follows:

   -  The software is successfully migrated from one version to a
      successive version or vice versa.

   -  There are no control-plane interruptions throughout the process.
      That is, the upgrade/downgrade could be accomplished while the
      device remains "in service".  It is noted, however, that most
      service providers will still undertake such actions in a
      maintenance window (even in redundant environments) to minimize
      any risk.

   -  Interruptions to the forwarding plane are minimal to none.

   -  The total time to accomplish the upgrade is minimized, again to
      reduce potential network outage exposure (e.g., an external
      failure event might impact the network as it operates with reduced

   This document provides a set of procedures to characterize a given
   forwarding device's ISSU behavior quantitatively, from the
   perspective of meeting the above expectations.

   Different hardware configurations may be expected to be benchmarked,
   but a typical configuration for a forwarding device that supports
   ISSU consists of at least one pair of Routing Processors (RPs) that
   operate in a redundant fashion, and single or multiple forwarding
   engines (line cards) that may or may not be redundant, as well as
   fabric cards or other components as applicable.  This does not
   preclude the possibility that a device in question can perform ISSU
   functions through the operation of independent process components,

Banks, et al.                 Informational                     [Page 3]

RFC 7654              Benchmarking Software Upgrade         October 2015

   which may be upgraded without impact to the overall operation of the
   device.  As an example, perhaps the software module involved in SNMP
   functions can be upgraded without impacting other operations.

   The concept of a multi-chassis deployment may also be characterized
   by the current set of proposed methodologies, but the implementation-
   specific details (i.e., process placement and others) are beyond the
   scope of the current document.

   Since most modern forwarding devices, where ISSU would be applicable,
   do consist of redundant RPs and hardware-separated control-plane and
   data-plane functionality, this document will focus on methodologies
   that would be directly applicable to those platforms.  It is
   anticipated that the concepts and approaches described herein may be
   readily extended to accommodate other device architectures as well.

2.  Conventions Used in This Document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   document are to be interpreted as described in RFC 2119 [RFC2119].

   In this document, these words will appear with that interpretation
   only when in ALL CAPS.  Lowercase uses of these words are not to be
   interpreted as carrying the significance of RFC 2119.

3.  Generic ISSU Process, Phased Approach

   ISSU may be viewed as the behavior of a device when exposed to a
   planned change in its software functionality.  This may mean changes
   to the core operating system, separate processes or daemons, or even
   firmware logic in programmable hardware devices (e.g., Complex
   Programmable Logic Device (CPLD) or Field-Programmable Gate Array
   (FPGA)).  The goal of an ISSU implementation is to permit such
   actions with minimal or no disruption to the primary operation of the
   device in question.

   ISSU may be user initiated through direct interaction with the device
   or activated through some automated process on a management system or
   even on the device itself.  For the purposes of this document, we
   will focus on the model where the ISSU action is initiated by direct
   user intervention.

   The ISSU process can be viewed as a series of different phases or
   activities, as defined below.  For each of these phases, the test
   operator must record the outcome as well as any relevant observations
   (defined further in the present document).  Note that, a given vendor
   implementation may or may not permit the abortion of the in-progress

Banks, et al.                 Informational                     [Page 4]

RFC 7654              Benchmarking Software Upgrade         October 2015

   ISSU at particular stages.  There may also be certain restrictions as
   to ISSU availability given certain functional configurations (for
   example, ISSU in the presence of Bidirectional Failure Detection
   (BFD) [RFC5880] may not be supported).  It is incumbent upon the test
   operator to ensure that the DUT is appropriately configured to
   provide the appropriate test environment.  As with any properly
   orchestrated test effort, the test plan document should reflect these
   and other relevant details and should be written with close attention
   to the expected production operating environment.  The combined
   analysis of the results of each phase will characterize the overall
   ISSU process with the main goal of being able to identify and
   quantify any disruption in service (from the data- and control-plane
   perspective) allowing operators to plan their maintenance activities
   with greater precision.

3.1.  Software Download

   In this first phase, the requested software package may be downloaded
   to the router and is typically stored onto a device.  The downloading
   of software may be performed automatically by the device as part of
   the upgrade process, or it may be initiated separately.  Such
   separation allows an administrator to download the new code inside or
   outside of a maintenance window; it is anticipated that downloading
   new code and saving it to disk on the router will not impact
   operations.  In the case where the software can be downloaded outside
   of the actual upgrade process, the administrator should do so;
   downloading software can skew timing results based on factors that
   are often not comparative in nature.  Internal compatibility
   verification may be performed by the software running on the DUT, to
   verify the checksum of the files downloaded as well as any other
   pertinent checks.  Depending upon vendor implementation, these
   mechanisms may include 1) verifying that the downloaded module(s)
   meet a set of identified prerequisites such as (but not limited to)
   hardware or firmware compatibility or minimum software requirements
   or even 2) ensuring that device is "authorized" to run the target

   Where such mechanisms are made available by the product, they should
   be verified, by the tester, with the goal of avoiding operational
   issues in production.  Verification should include both positive
   verification (ensuring that an ISSU action should be permitted) as
   well as negative tests (creation of scenarios where the verification
   mechanisms would report exceptions).

Banks, et al.                 Informational                     [Page 5]

RFC 7654              Benchmarking Software Upgrade         October 2015

3.2.  Software Staging

   In this second phase, the requested software package is loaded in the
   pertinent components of a given forwarding device (typically the RP
   in standby state).  Internal compatibility verification may be
   performed by the software running on the DUT, as part of the upgrade
   process itself, to verify the checksum of the files downloaded as
   well as any other pertinent checks.  Depending upon vendor
   implementation, these mechanisms may include verification that the
   downloaded module(s) meet a set of identified prerequisites such as
   hardware or firmware compatibility or minimum software requirements.
   Where such mechanisms are made available by the product, they should
   be verified, by the tester (again with the goal of avoiding
   operational issues in production).  In this case, the execution of
   these checks is within the scope of the upgrade time and should be
   included in the testing results.  Once the new software is downloaded
   to the pertinent components of the DUT, the upgrade begins, and the
   DUT begins to prepare itself for upgrade.  Depending on the vendor
   implementation, it is expected that redundant hardware pieces within
   the DUT are upgraded, including the backup or secondary RP.

3.3.  Upgrade Run

   In this phase, a switchover of RPs may take place, where one RP is
   now upgraded with the new version of software.  More importantly, the
   "Upgrade Run" phase is where the internal changes made to information
   and state (stored on the router, on disk, and in memory) are either
   migrated to the "new" version of code, or transformed/rebuilt to meet
   the standards of the new version of code, and pushed onto the
   appropriate pieces of hardware.  It is within this phase that any
   outage(s) on the control or forwarding plane may be expected to be
   observed.  This is the critical phase of the ISSU, where the control
   plane should not be impacted and any interruptions to the forwarding
   plane should be minimal to none.

   If any control- or data-plane interruptions are observed within this
   stage, they should be recorded as part of the results document.

   For some implementations, the two stages, as described in Section 3.2
   and above, may be concatenated into one monolithic operation.  In
   that case, the calculation of the respective ISSU time intervals may
   need to be adapted accordingly.

Banks, et al.                 Informational                     [Page 6]

RFC 7654              Benchmarking Software Upgrade         October 2015

3.4.  Upgrade Acceptance

   In this phase, the new version of software must be running in all the
   physical nodes of the logical forwarding device (RPs and line cards
   as applicable).  At this point, configuration control is returned to
   the operator, and normal device operation, i.e., outside of ISSU-
   oriented operation, is resumed.

4.  Test Methodology

   As stated by [RFC6815], the Test Topology Setup must be part of an
   Isolated Test Environment (ITE).

   The reporting of results must take into account the repeatability
   considerations from Section 4 of [RFC2544].  It is RECOMMENDED to
   perform multiple trials and report average results.  The results are
   reported in a simple statement including the measured frame loss and
   ISSU impact times.

4.1.  Test Topology

   The hardware configuration of the DUT (Device Under Test) should be
   identical to the one expected to be or currently deployed in
   production in order for the benchmark to have relevance.  This would
   include the number of RPs, hardware version, memory, and initial
   software release, any common chassis components, such as fabric
   hardware in the case of a fabric-switching platform, and the specific
   line cards (version, memory, interfaces type, rate, etc.).

   For the control and data plane, differing configuration approaches
   may be utilized.  The recommended approach relies on "mimicking" the
   existing production data- and control-plane information, in order to
   emulate all the necessary Layer 1 through Layer 3 communications and,
   if appropriate, the upper-layer characteristics of the network, as
   well as end-to-end traffic/communication pairs.  In other words,
   design a representative load model of the production environment and
   deploy a collapsed topology utilizing test tools and/or external
   devices, where the DUT will be tested.  Note that, the negative
   impact of ISSU operations is likely to impact scaled, dynamic
   topologies to a greater extent than simpler, static environments.  As
   such, this methodology (based upon production configuration) is
   advised for most test scenarios.

   The second, more simplistic approach is to deploy an ITE in which
   endpoints are "directly" connected to the DUT.  In this manner,
   control-plane information is kept to a minimum (only connected
   interfaces), and only a basic data-plane of sources and destinations
   is applied.  If this methodology is selected, care must be taken to

Banks, et al.                 Informational                     [Page 7]

RFC 7654              Benchmarking Software Upgrade         October 2015

   understand that the systemic behavior of the ITE may not be identical
   to that experienced by a device in a production network role.  That
   is, control-plane validation may be minimal to none with this
   methodology.  Consequently, if this approach is chosen, comparison
   with at least one production configuration is recommended in order to
   understand the direct relevance and limitations of the test exercise.

4.2.  Load Model

   In consideration of the defined test topology, a load model must be
   developed to exercise the DUT while the ISSU event is introduced.
   This applied load should be defined in such a manner as to provide a
   granular, repeatable verification of the ISSU impact on transit
   traffic.  Sufficient traffic load (rate) should be applied to permit
   timing extrapolations at a minimum granularity of 100 milliseconds,
   e.g., 100 Mbps for a 10 Gbps interface.  The use of steady traffic
   streams rather than bursty loads is preferred to simplify analysis.

   The traffic should be patterned to provide a broad range of source
   and destination pairs, which resolve to a variety of FIB (Forwarding
   Information Base) prefix lengths.  If the production network
   environment includes multicast traffic or VPNs (L2, L3, or IPsec), it
   is critical to include these in the model.

   For mixed protocol environments (e.g., IPv4 and IPv6), frames should
   be distributed between the different protocols.  The distribution
   should approximate the network conditions of deployment.  In all
   cases, the details of the mixed protocol distribution must be
   included in the reporting.

   The feature, protocol timing, and other relevant configurations
   should be matched to the expected production environment.  Deviations
   from the production templates may be deemed necessary by the test
   operator (for example, certain features may not support ISSU or the
   test bed may not be able to accommodate such).  However, the impact
   of any such divergence should be clearly understood, and the
   differences must be recorded in the results documentation.  It is
   recommended that a Network Management System (NMS) be deployed,
   preferably similar to that utilized in production.  This will allow
   for monitoring of the DUT while it is being tested, both in terms of
   supporting the impact analysis on system resources as well as
   detecting interference with non-transit (management) traffic as a
   result of the ISSU operation.  It is suggested that the actual test
   exercise be managed utilizing direct console access to the DUT, if at
   all possible, to avoid the possibility that a network interruption
   impairs execution of the test exercise.

Banks, et al.                 Informational                     [Page 8]

RFC 7654              Benchmarking Software Upgrade         October 2015

   All in all, the load model should attempt to simulate the production
   network environment to the greatest extent possible in order to
   maximize the applicability of the results generated.

5.  ISSU Test Methodology

   As previously described, for the purposes of this test document, the
   ISSU process is divided into three main phases.  The following
   methodology assumes that a suitable test topology has been
   constructed per Section 4.  A description of the methodology to be
   applied for each of the above phases follows.

5.1.  Pre-ISSU Recommended Verifications

   The steps of this phase are as follows.

   1.  Verify that enough hardware and software resources are available
       to complete the Load operation (e.g., enough disk space).

   2.  Verify that the redundancy states between RPs and other nodes are
       as expected (e.g., redundancy on, RPs synchronized).

   3.  Verify that the device, if running protocols capable of NSR (Non-
       Stop Routing), is in a "ready" state; that is, that the sync
       between RPs is complete and the system is ready for failover, if

   4.  Gather a configuration snapshot of the device and all of its
       applicable components.

   5.  Verify that the node is operating in a "steady" state (that is,
       no critical or maintenance function is being currently

   6.  Note any other operational characteristics that the tester may
       deem applicable to the specific implementation deployed.

5.2.  Software Staging

   The steps of this phase are as follows.

   1.  Establish all relevant protocol adjacencies and stabilize routing
       within the test topology.  In particular, ensure that the scaled
       levels of the dynamic protocols are dimensioned as specified by
       the test topology plan.

Banks, et al.                 Informational                     [Page 9]

RFC 7654              Benchmarking Software Upgrade         October 2015

   2.  Clear, relevant logs and interface counters to simplify analysis.
       If possible, set logging timestamps to a highly granular mode.
       If the topology includes management systems, ensure that the
       appropriate polling levels have been applied, sessions have been
       established, and the responses are per expectation.

   3.  Apply the traffic loads as specified in the load model previously
       developed for this exercise.

   4.  Document an operational baseline for the test bed with relevant
       data supporting the above steps (include all relevant load
       characteristics of interest in the topology, e.g., routing load,
       traffic volumes, memory and CPU utilization).

   5.  Note the start time (T0) and begin the code change process
       utilizing the appropriate mechanisms as expected to be used in
       production (e.g., active download with TFTP, FTP, SCP, etc., or
       direct install from local or external storage facility).  In
       order to ensure that ISSU process timings are not skewed by the
       lack of a network-wide synchronization source, the use of a
       network NTP source is encouraged.

   6.  Take note of any logging information and command-line interface
       (CLI) prompts as needed.  (This detail will be vendor specific.)
       Respond to any DUT prompts in a timely manner.

   7.  Monitor the DUT for the reload of the secondary RP to the new
       software level.  Once the secondary has stabilized on the new
       code, note the completion time.  The duration of these steps will
       be recorded as "T1".

   8.  Review system logs for any anomalies, check that relevant dynamic
       protocols have remained stable, and note traffic loss if any.
       Verify that deployed management systems have not identified any
       unexpected behavior.

5.3.  Upgrade Run

   The following assumes that the software load step and upgrade step
   are discretely controllable.  If not, maintain the aforementioned
   timer and monitor for completion of the ISSU as described below.

   1.  Note the start time and initiate the actual upgrade procedure.

   2.  Monitor the operation of the secondary route processor while it
       initializes with the new software and assumes mastership of the
       DUT.  At this point, pay particular attention to any indications
       of control-plane disruption, traffic impact, or other anomalous

Banks, et al.                 Informational                    [Page 10]

RFC 7654              Benchmarking Software Upgrade         October 2015

       behavior.  Once the DUT has converged upon the new code and
       returned to normal operation, note the completion time and log
       the duration of this step as "T2".

   3.  Review the syslog data in the DUT and neighboring devices for any
       behavior that would be disruptive in a production environment
       (line card reloads, control-plane flaps, etc.).  Examine the
       traffic generators for any indication of traffic loss over this
       interval.  If the Test Set reported any traffic loss, note the
       number of frames lost as "TPL_frames", where TPL stands for
       "Total Packet Loss".  If the Test Set also provides outage
       duration, note this as "TPL_time".  (Alternatively, TPL_time may
       be calculated as (TPL / Offered Load) * 1000.  The units for
       Offered Load are packets per second; the units for TPL_time are

   4.  Verify the DUT status observations as per any NMS managing the
       DUT and its neighboring devices.  Document the observed CPU and
       memory statistics both during and after the ISSU upgrade event,
       and ensure that memory and CPU have returned to an expected
       (previously baselined) level.

5.4.  Post-ISSU Verification

   The following describes a set of post-ISSU verification tasks that
   are not directly part of the ISSU process, but are recommended for
   execution in order to validate a successful upgrade.

   1.  Configuration delta analysis

       Examine the post-ISSU configurations to determine if any changes
       have occurred either through process error or due to differences
       in the implementation of the upgraded code.

   2.  Exhaustive control-plane analysis

       Review the details of the Routing Information Base (RIB) and FIB
       to assess whether any unexpected changes have been introduced in
       the forwarding paths.

   3.  Verify that both RPs are up and that the redundancy mechanism for
       the control plane is enabled and fully synchronized.

   4.  Verify that no control-plane (protocol) events or flaps were

   5.  Verify that no L1 and or L2 interface flaps were observed.

Banks, et al.                 Informational                    [Page 11]

RFC 7654              Benchmarking Software Upgrade         October 2015

   6.  Document the hitless operation or presence of an outage based
       upon the counter values provided by the Test Set.

5.5.  ISSU under Negative Stimuli

   As an OPTIONAL Test Case, the operator may want to perform an ISSU
   test while the DUT is under stress by introducing route churn to any
   or all of the involved phases of the ISSU process.

   One approach relies on the operator to gather statistical information
   from the production environment and determine a specific number of
   routes to flap every 'fixed' or 'variable' interval.  Alternatively,
   the operator may wish to simply preselect a fixed number of prefixes
   to flap.  As an example, an operator may decide to flap 1% of all the
   BGP routes every minute and restore them 1 minute afterwards.  The
   tester may wish to apply this negative stimulus throughout the entire
   ISSU process or, most importantly, during the run phase.  It is
   important to ensure that these routes, which are introduced solely
   for stress proposes, must not overlap the ones (per the load model)
   specifically leveraged to calculate the TPL_time (recorded outage).
   Furthermore, there should not be 'operator-induced' control-plane
   protocol adjacency flaps for the duration of the test process as it
   may adversely affect the characterization of the entire test
   exercise.  For example, triggering IGP adjacency events may force
   recomputation of underlying routing tables with attendant impact to
   the perceived ISSU timings.  While not recommended, if such trigger
   events are desired by the test operator, care should be taken to
   avoid the introduction of unexpected anomalies within the test

6.  ISSU Abort and Rollback

   Where a vendor provides such support, the ISSU process could be
   aborted for any reason by the operator.  However, the end results and
   behavior may depend on the specific phase where the process was
   aborted.  While this is implementation dependent, as a general
   recommendation, if the process is aborted during the "Software
   Download" or "Software Staging" phases, no impact to service or
   device functionality should be observed.  In contrast, if the process
   is aborted during the "Upgrade Run" or "Upgrade Accept" phases, the
   system may reload and revert back to the previous software release,
   and, as such, this operation may be service affecting.  Where vendor
   support is available, the abort/rollback functionality should be
   verified, and the impact, if any, quantified generally following the
   procedures provided above.

Banks, et al.                 Informational                    [Page 12]

RFC 7654              Benchmarking Software Upgrade         October 2015

7.  Final Report: Data Presentation and Analysis

   All ISSU impact results are summarized in a simple statement
   describing the "ISSU Disruption Impact" including the measured frame
   loss and impact time, where impact time is defined as the time frame
   determined per the TPL_time reported outage.  These are considered to
   be the primary data points of interest.

   However, the entire ISSU operational impact should also be considered
   in support of planning for maintenance, and, as such, additional
   reporting points are included.

      Software download / secondary update      T1

      Upgrade/Run                               T2

      ISSU Traffic Disruption (Frame Loss)      TPL_frames

      ISSU Traffic Impact Time (milliseconds)   TPL_time

      ISSU Housekeeping Interval                T3

      (Time for both RPs up on new code and fully synced - Redundancy

      Total ISSU Maintenance Window             T4 (sum of T1+T2+T3)

   The results reporting must provide the following information:

   -  DUT hardware and software detail

   -  Test Topology definition and diagram (especially as related to the
      ISSU operation)

   -  Load Model description including protocol mixes and any divergence
      from the production environment

   -  Time Results as per above

   -  Anomalies Observed during ISSU

   -  Anomalies Observed in post-ISSU analysis

Banks, et al.                 Informational                    [Page 13]

RFC 7654              Benchmarking Software Upgrade         October 2015

   It is RECOMMENDED that the following parameters be reported as
   outlined below:

   Parameter                Units or Examples

   Traffic Load             Frames per second and bits per second

   Disruption (average)     Frames

   Impact Time (average)    Milliseconds

   Number of trials         Integer count

   Protocols                IPv4, IPv6, MPLS, etc.

   Frame Size               Octets

   Port Media               Ethernet, Gigabit Ethernet (GbE),
                            Packet over SONET (POS), etc.

   Port Speed               10 Gbps, 1 Gbps, 100 Mbps, etc.

   Interface Encaps         Ethernet, Ethernet VLAN, PPP,
                            High-Level Data Link Control (HDLC), etc.

   Number of Prefixes       Integer count

   flapped (ON Interval)    (Optional)  # of prefixes / Time (min.)

   flapped (OFF Interval)   (Optional)  # of prefixes / Time (min.)

   Document any configuration deltas that are observed after the ISSU
   upgrade has taken effect.  Note differences that are driven by
   changes in the patch or release level, as well as items that are
   aberrant changes due to software faults.  In either of these cases,
   any unexpected behavioral changes should be analyzed and a
   determination made as to the impact of the change (be it functional
   variances or operational impacts to existing scripts or management

7.1.  Data Collection Considerations

   When a DUT is undergoing an ISSU operation, it's worth noting that
   the DUT's data collection and reporting of data, such as counters,
   interface statistics, log messages, etc., may not be accurate.  As
   such, one should not rely on the DUT's data collection methods, but
   rather, should use the test tools and equipment to collect data used

Banks, et al.                 Informational                    [Page 14]

RFC 7654              Benchmarking Software Upgrade         October 2015

   for reporting in Section 7.  Care and consideration should be paid in
   testing or adding new test cases, such that the desired data can be
   collected from the test tools themselves, or other external
   equipment, outside of the DUT itself.

8.  Security Considerations

   All BMWG memos are limited to testing in a laboratory Isolated Test
   Environment (ITE), thus avoiding accidental interruption to
   production networks due to test activities.

   All benchmarking activities are limited to technology
   characterization using controlled stimuli in a laboratory environment
   with dedicated address space and the other constraints [RFC2544].

   The benchmarking network topology will be an independent test setup
   and MUST NOT be connected to devices that may forward the test
   traffic into a production network or misroute traffic to the test
   management network.

   Further, benchmarking is performed on a "black-box" basis, relying
   solely on measurements observable external to the Device Under Test /
   System Under Test (DUT/SUT).

   Special capabilities should not exist in the DUT/SUT specifically for
   benchmarking purposes.  Any implications for network security arising
   from the DUT/SUT should be identical in the lab and in production

9.  References

9.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,

   [RFC2544]  Bradner, S. and J. McQuaid, "Benchmarking Methodology for
              Network Interconnect Devices", RFC 2544,
              DOI 10.17487/RFC2544, March 1999,

Banks, et al.                 Informational                    [Page 15]

RFC 7654              Benchmarking Software Upgrade         October 2015

9.2.  Informative References

   [RFC5880]  Katz, D. and D. Ward, "Bidirectional Forwarding Detection
              (BFD)", RFC 5880, DOI 10.17487/RFC5880, June 2010,

   [RFC6815]  Bradner, S., Dubray, K., McQuaid, J., and A. Morton,
              "Applicability Statement for RFC 2544: Use on Production
              Networks Considered Harmful", RFC 6815,
              DOI 10.17487/RFC6815, November 2012,


   The authors wish to thank Vibin Thomas for his valued review and

Authors' Addresses

   Sarah Banks
   VSS Monitoring
   Email: sbanks@encrypted.net

   Fernando Calabria
   Cisco Systems
   Email: fcalabri@cisco.com

   Gery Czirjak
   Juniper Networks
   Email: gczirjak@juniper.net

   Ramdas Machat
   Juniper Networks
   Email: rmachat@juniper.net

Banks, et al.                 Informational                    [Page 16]