Cloud computing - Private Cloud

Cloud computing is the new style of computing that is scalable and use virtualized resource through
internet. The scalability of cloud computing give benefits to big data because data warehouse can be
scaled up when it is required to be, and could be shrink unless it is required
(Sharda, Delen & Turban 2018).

Figure. conceptual architecture of a cloud-oriented system (Sharda, Delen & Turban 2018).

Private cloud called internal cloud is more secure than the public cloud such as MS Azure or 

BigQuery of Google. It has the same benefits of public cloud in terms of scalability, changing the
resource according to the demand, etc. On top of this advantage, private cloud also resolve the
security issue of cloud system by having direct control by EROAD over the data stored in the cloud.

One disadvantage of private cloud is the cost that is required for the management of cloud system
by the company (Sharda, Delen & Turban 2018).

Sharda, R, Delen, D & Turban, E 2018, Business intelligence, analytics, and data science:
a managerial perspective, Pearson, Harlow, England.

Compare 802.11n VS 802.11ac

802.11n use both 2.4Ghz and 5Ghz while 802.11ac only use 5Ghz service band. 802.11n support 100 to 600 Mbps while 802.11ac support 433Mbps to 6.93Gbps.The channel bandwidth of 802.11n is 40Mhz and can be fall back to 20Mhz in case of interference, while the channel bandwidth of 802.11ac is 80Mhz and 160Mhz. In 5Ghz service band, 802.11n can have 20 ~25 20Mhz channels while 802.11ac can have 4 ~6 80Mhz or 1~2 160Mhz channels. The maximum spatial streams in 802.11n are 4 while, the maximum spatial streams in 802.11ac is 8. Multiuser MIMO is not possible in 802.11n while it is possible with 802.11ac.

Compare 2.4Ghz vs 5Ghz

2.4 Ghz has 3 non-overlapping 20Mhz width channel while 5Ghz has 11 to 24 * 20Mhz non-overlapping channel. So, the possibility of co-channel interference in 2.4Ghz is higher than the possibility in 5Ghz, and more bandwidth could be provided in 5Ghz by installing more AP in small area while avoiding co-channel interference.

With 802.11n, 40Mhz width could be used in 2.4Ghz, but it will be shrink to 20Mhz bandwidth if there is co-channel interference. With 802.11N and 802.11AC, up to 160Mhz channel width could be used in 5 Ghz.

However, the attenuation rate in 5Ghz is higher than the one in 2.4 Ghz because of high level of attenuation in higher frequency by obstacle or absorptive attenuation by air or water.So, the coverage by ap running in 2.4Ghz is wider than the coverage of AP running in 5Ghz, and 2.4Ghz is appropriate for the place where require wider range of coverage with less number of user and APs.

Python script to download and install EXOS image multiple times

Below Python script and UPM configuration could be used to try upgrade on the switch multiple times.
This script used to make switch download and install EXOS image multiple times.

[upgrade.py]

import time

import re

import sys

def upgrade():

for loop in range(1, 51):

upgradeCMD = 'download image x.x.x.x summitX-16.2.3.5-patch1-3.xos vr
"VR-Mgmt" primary'

result = open('/usr/local/tmp/upgraderesult.txt', 'a')

cmdOutput = exsh.clicmd(upgradeCMD,capture=True)

result.write(cmdOutput)

time.sleep(10)

result.close()

if __name__ == '__main__':

   try:

       upgrade()

   except SystemExit:

       pass

[configuration]

create upm profile testprofile

load script upgrade.py

.

configure upm profile testprofile maximum execution-time 590

create upm timer testtimer

configure upm timer testtimer profile testprofile

configure upm timer testtimer after 10 every 900

TCL script to download and install EXOS multiple times

This is an UPM script using TCL and configuration on EXOS to trigger this UPM script. The purpose of this script is upgrading EXOS repeatedly to see whether there is any issue during EXOS upgrade.

[ upgrade.xsf ]

set var temp $TCL(after [expr 60 * 1000])

set var cli.out " "

show switch | in Image

set var output $TCL(split ${cli.out} "\n")

set var line $TCL(lindex $output 1)

set var split0 $TCL(split $line ":")

set var partition $TCL(lindex $split0 1)

set var TF $TCL(lsearch $partition *primary* )

if ($TF) then

 download image 10.67.55.52 summitX-15.6.4.2-patch1-6.xos vr "VR-Mgmt" primary

elseif

 download image 10.67.55.52 summitX-15.6.4.2-patch1-6.xos vr "VR-Mgmt" secondary

endif

reboot

[ config ]

create log filter testfilter

configure log filter testfilter add events HAL.Card.Info match string "Switch is operational"

create log target upm test

configure log target upm test filter testfilter severity Info

enable log target upm

create upm profile test

load script upgrade.xsf

.

configure upm profile test maximum execution-time 600

configure vlan mgmt ipaddress y.y.y.y

configure syslog add x.x.x.x:514 vr VR-Mgmt local1

enable log target syslog x.x.x.x:514 vr VR-Mgmt local1

Troubleshooting OSPF neighbor flapping by TCL script on EXOS

If OSPF problem is existing in customer’s network, we could narrow down the cause of this issue by the log filter and access-list, etc.

( whether ospf hello message dropped on sender side or receiver side, if it is receiver side then, whether is dropped on slow patch or it is not found on HW etc )

Please see below example and result in the lab.

1. log filter

[config ]

configure log filter DefaultFilter add events ospf.hello.PktRx match ipaddress "1.1.1.2"

[result ]

* cntcom01.75 # sh log

11/21/2017 13:44:46.89 <Info:ospf.hello.PktRx> MSM-A: test received a Hello pkt from 1.1.1.2

11/21/2017 13:44:36.88 <Info:ospf.hello.PktRx> MSM-A: test received a Hello pkt from 1.1.1.2

2. access-list

Policy: ospf1

entry e1 {

if match all {

   source-address 1.1.1.2/32 ;

   destination-address 224.0.0.5/32 ;

}

then {

   count ospfc_1.1.1.2 ;

}

}

entry e2 {

if match all {

   source-address 1.1.1.3/32 ;

   destination-address 224.0.0.5/32 ;

}

then {

   count ospfc_1.1.1.3 ;

}

}

configure access-list ospf1 vlan test

3. UPM

[ ospfcount.xsf ]

set var cli.out " "

show access-list counter

clear access-list counter

set var output $TCL(split ${cli.out} "\n")

set var len $TCL(llength $output)

set var total ($len - 6 )

set var cnt 0

set var ospflineindex $TCL(lsearch $output *ospfc* )

while ($cnt < $total) do

           set var line $TCL(lindex $output $ospflineindex)

           set var ospfcount $TCL(lindex $line 1)

           create log entry OSPF_COUNT:$ospfcount

           set var cnt ($cnt + 1)

           set var ospflineindex ($ospflineindex +1)

endwhile

[config on switch]

create upm profile ospfcount

load script ospfcount.xsf

.

create upm timer ospfcount_timer

configure upm timer ospfcount_timer profile ospfcount

configure upm timer ospfcount_timer after 10 every 60

[result]

11/21/2017 14:46:34.99 <Info:System.userComment> OSPF_COUNT:6

11/21/2017 14:46:34.97 <Info:System.userComment> OSPF_COUNT:6

11/21/2017 14:46:34.86 <Noti:UPM.Msg.upmMsgExshLaunch> Launched profile ospfcount for the event UPM Timer ospfcount_timer

11/21/2017 14:45:34.35 <Info:System.userComment> OSPF_COUNT:6

11/21/2017 14:45:34.33 <Info:System.userComment> OSPF_COUNT:6

11/21/2017 14:45:34.20 <Noti:UPM.Msg.upmMsgExshLaunch> Launched profile ospfcount for the event UPM Timer ospfcount_timer

4. on neighbor switch ( for example, on 202.251.2.48 )

configure log filter DefaultFilter add events ospf.hello.TxPkt

enable log debug-mode

[result]

* cntcos01.77 # sh log

11/21/2017 14:49:57.24 <Summ:ospf.hello.TxPkt> MSM-A: test is transmitting a Hello pkt to 224.0.0.5 (length 56, interface 0x0050B638, pkt num 8293).

11/21/2017 14:49:47.24 <Summ:ospf.hello.TxPkt> MSM-A: test is transmitting a Hello pkt to 224.0.0.5 (length 56, interface 0x0050B638, pkt num 8292).

How Qos Work for untagged traffic in EXOS based switch.

In EXOS, qos based on dot1p value have priority than DSCP value. However, in case of untagged traffic, it does not have tag, so does not have dot1p value.

The QOS logic for those traffic is different per EXOS version.

[Diagram]

Before EXOS version 16.1, it use DSCP value for egress QP.

Slot-1 Stack.3 # show diffserv examination

CodePoint-> QOSProfile mapping:

    00->QP1 01->QP1 02->QP1 03->QP1 04->QP1 05->QP1 06->QP1 07->QP1

    08*>QP1 09*>QP1 10*>QP1 11*>QP1 12*>QP1 13*>QP1 14*>QP1 15*>QP1

    16*>QP1 17*>QP1 18*>QP1 19*>QP1 20*>QP1 21*>QP1 22*>QP1 23*>QP1

    24*>QP1 25*>QP1 26*>QP1 27*>QP1 28*>QP1 29*>QP1 30*>QP1 31*>QP1

    32*>QP1 33*>QP1 34*>QP1 35*>QP1 36*>QP1 37*>QP1 38*>QP1 39*>QP1

    40*>QP1 41*>QP1 42*>QP1 43*>QP1 44*>QP1 45*>QP1 46*>QP1 47*>QP1

    48*>QP1 49*>QP1 50*>QP1 51*>QP1 52*>QP1 53*>QP1 54*>QP1 55*>QP1

    56*>QP5 57->QP8 58->QP8 59->QP8 60->QP8 61->QP8 62->QP8 63->QP8

Slot-1 Stack.5 # show ports 1:1 qosmonitor no-refresh

Port Qos Monitor

Port   QP1      QP2   QP3     QP4   QP5    QP6   QP7   QP8

   Pkt      Pkt   Pkt     Pkt   Pkt    Pkt   Pkt   Pkt

  Xmts     Xmts Xmts    Xmts Xmts   Xmts Xmts Xmts

===========================================================================

1:1 0        0     0       0  88325086   0      0    50

After EXOS version 16.1, it use default dot1p value configured on the port by below commands.

configure ports <ports> dot1p <dot1p_priority>

As a result, traffic with high DSCP ( for example, DSCP 56 also use lowest QP -QP1 )

* Slot-1 Stack.7 # show diffserv examination

CodePoint-> QOSProfile mapping:

       00->QP1 01->QP1 02->QP1 03->QP1 04->QP1 05->QP1 06->QP1 07->QP1

       08*>QP1 09*>QP1 10*>QP1 11*>QP1 12*>QP1 13*>QP1 14*>QP1 15*>QP1

       16*>QP1 17*>QP1 18*>QP1 19*>QP1 20*>QP1 21*>QP1 22*>QP1 23*>QP1

       24*>QP1 25*>QP1 26*>QP1 27*>QP1 28*>QP1 29*>QP1 30*>QP1 31*>QP1

       32*>QP1 33*>QP1 34*>QP1 35*>QP1 36*>QP1 37*>QP1 38*>QP1 39*>QP1

       40*>QP1 41*>QP1 42*>QP1 43*>QP1 44*>QP1 45*>QP1 46*>QP1 47*>QP1

       48*>QP1 49*>QP1 50*>QP1 51*>QP1 52*>QP1 53*>QP1 54*>QP1 55*>QP1

       56->QP8 57->QP8 58->QP8 59->QP8 60->QP8 61->QP8 62->QP8 63->QP8

* Slot-1 Stack.11 # show ports 1:1 qosmonitor no-refresh

Port Qos Monitor

Port           QP1        QP2        QP3        QP4     QP5        QP6        QP7     QP8

              Pkt        Pkt        Pkt        Pkt     Pkt        Pkt     Pkt     Pkt

             Xmts       Xmts       Xmts       Xmts    Xmts       Xmts       Xmts    Xmts

==========================================================================================

1:1     2734132554       0          0        0        0       0          0    76048