This document discusses calculating buffer credit requirements for long distance fibre channel links. It provides an overview of fibre channel frame structure and sizes. It then shows calculations for link distances and speeds to determine the number of frames and bits that could be in transit simultaneously. This is used to calculate the required number of buffer-to-buffer credits. It proposes a rule of thumb that for a 4Gbps link, 1.33 buffer credits are needed per kilometer. It notes to add an additional 20% credits as recommended by best practices. The document seeks feedback on the proposed 1.33 factor for 4Gbps links.
Here is the presentation for Physical Layer Numericals from the book Andrew S. Tanenbaum (Computer Networks) and B A Forouzan ( Data Communication and Networking)
Here is the presentation for Physical Layer Numericals from the book Andrew S. Tanenbaum (Computer Networks) and B A Forouzan ( Data Communication and Networking)
Telecommunication System Engineering NotesHaris Hassan
Telecommunications engineering, or telecoms engineering, is an engineering discipline centered on electrical and computer engineering which seeks to support and enhance telecommunication systems.
Telecommunication System Engineering NotesHaris Hassan
Telecommunications engineering, or telecoms engineering, is an engineering discipline centered on electrical and computer engineering which seeks to support and enhance telecommunication systems.
Engineer EMERSON EDUARDO RODRIGUES PRESENTA UNA NUEVA VERSION
THERE ONE NEW ONE PRESENTATION FOR 2G AND 3G ENGINEERING FOR LTE AND PSCORE ENGINEER
ITS VERY SUITABLE FOR YOUR RESEARCH AT ALL LEVELS OF RF ENGINEERING AND PS CS
A Carrierless Amplitude Phase (CAP) Modulation Format: Perspective and Prospe...IJECEIAES
The explosive demand of broadband services nowadays requires data communication systems to have intensive capacity which subsequently increases the need for higher data rate as well. Although implementation of multiple wavelengths channels can be used (e.g. 4 × 25.8 Gb/s for 100 Gb/s connection) for such desired system, it usually leads to cost increment issue which is caused by employment of multiple optical components. Therefore, implementation of advanced modulation format using a single wavelength channel has become a preference to increase spectral efficiency by increasing the data rate for a given transmission system bandwidth. Conventional advanced modulation format however, involves a degree of complexity and costly transmission system. Hence, carrierless amplitude phase (CAP) modulation format has emerged as a promising advanced modulation format candidate due to spectral efficiency improvement ability with reduction of optical transceiver complexity and cost. The intriguing properties of CAP modulation format are reviewed as an attractive prospect in optical transmission system applications.
Focusing on the physical layer of the 5G network, new methodologies, new problems facing the network and solutions for each problem.
topics are:-
Channel Models, Channel Coding, Multiple Access, Smart Antenna, Massive MIMO & Beamforming, Network Architecture, Frame structure & Numerology
addition: exploring the new trends that might be done in the future
MAT 510 – Homework AssignmentHomework Assignment 6 Due.docxjessiehampson
MAT 510 – Homework Assignment
Homework Assignment 6
Due in Week 9 and worth 30 points
Suppose the number of equipment sales and service contracts that a store sold during the last six (6) months for treadmills and exercise bikes was as follows:
Equipment Sales and Service Contracts Sold
Treadmill
Exercise Bike
Total Sold
185
123
Service Contracts
67
55
The store can only sell a service contract on a new piece of equipment. Of the 185 treadmills sold, 67 included a service contract and 118 did not.
Complete the following questions in the space provided below:
1. Construct a 95 percent confidence interval for the difference between the proportions of service contracts sold on treadmills versus exercise bikes.
2. Is there a major difference between the two pieces of equipment? Why or why not?
Type your answers below and submit this file in Week 9 of the online course shell:
802.11 THROUGHPUT
comp40660 Assignment 1, February 2020
This assignment is worth 18% of the overall grade
Motivation
• Build a simple model of 802.11 frame exchange for TCP
and UDP, using OFDM of 802.11a and 802.11g
• The model will approximate the actual throughput of the
network
• RTS/CTS mechanism is enabled
• No contention
• Demonstration of the calculation for 802.11a – UDP case;
work on TCP case in lab.
• Assignment will be to modify for the .11g/n/ac/ax case for
both TCP and UDP.
802.11 Model
• Basic transactional model – 2 different transaction types, namely
UDP and TCP.
• Any 802.11 transmission of data (from higher layer) requires an
acknowledgement (ACK) by the .11 MAC.
• Each TCP / UDP packet is encapsulated in a single 802.11 frame.
Transport
Network
Data Link
Physical
Transport
Network
Data Link
PhysicalBits
Frame
Packet
Segment
802.11 Frame Exchange
UDP Case
• No guarantee of delivery
• Suitable for real-time applications such as VoIP, VoD
• UDP data encapsulated into 802.11 frame and
transmitted. Receiving station transmits 802.11 ACK.
Server Client
UDP
802.11 Frame Exchange
TCP Case
• Reliable delivery service guaranteeing that all bytes are
received and in correct order through TCP ACKs
• How is this different from the UDP case?
TCP
ACK
Server Client
Data Transmission
• 802.11 uses different inter-frame spaces:
• SIFS (Short Interframe Space)
• High-priority transmissions can begin once SIFS has elapsed
• ACK, RTS, CTS
• DIFS (DCF Interframe Space)
• Minimum idle time for contention-based services
• Stations can have access to the medium if it has been free for
a period longer than DIFS
Packet Headers
• 1500 bytes packet (TCP/UDP) is encapsulated:
• MAC header = 34 bytes
• SNAP LLC header = 8 bytes
• 3 bytes LLC (logical link control) header
• 5 bytes SNAP (sub-network access protocol) header
=> Total size = 1542 bytes
802.11a
• Amendment to the IEEE 802.11 specification
• 1999
• 5Ghz band
• Maximum data rate: 54 Mbps
• OFDM (Orthogonal Frequency Division.
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...John Andrews
SlideShare Description for "Chatty Kathy - UNC Bootcamp Final Project Presentation"
Title: Chatty Kathy: Enhancing Physical Activity Among Older Adults
Description:
Discover how Chatty Kathy, an innovative project developed at the UNC Bootcamp, aims to tackle the challenge of low physical activity among older adults. Our AI-driven solution uses peer interaction to boost and sustain exercise levels, significantly improving health outcomes. This presentation covers our problem statement, the rationale behind Chatty Kathy, synthetic data and persona creation, model performance metrics, a visual demonstration of the project, and potential future developments. Join us for an insightful Q&A session to explore the potential of this groundbreaking project.
Project Team: Jay Requarth, Jana Avery, John Andrews, Dr. Dick Davis II, Nee Buntoum, Nam Yeongjin & Mat Nicholas
Explore our comprehensive data analysis project presentation on predicting product ad campaign performance. Learn how data-driven insights can optimize your marketing strategies and enhance campaign effectiveness. Perfect for professionals and students looking to understand the power of data analysis in advertising. for more details visit: https://bostoninstituteofanalytics.org/data-science-and-artificial-intelligence/
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Subhajit Sahu
Abstract — Levelwise PageRank is an alternative method of PageRank computation which decomposes the input graph into a directed acyclic block-graph of strongly connected components, and processes them in topological order, one level at a time. This enables calculation for ranks in a distributed fashion without per-iteration communication, unlike the standard method where all vertices are processed in each iteration. It however comes with a precondition of the absence of dead ends in the input graph. Here, the native non-distributed performance of Levelwise PageRank was compared against Monolithic PageRank on a CPU as well as a GPU. To ensure a fair comparison, Monolithic PageRank was also performed on a graph where vertices were split by components. Results indicate that Levelwise PageRank is about as fast as Monolithic PageRank on the CPU, but quite a bit slower on the GPU. Slowdown on the GPU is likely caused by a large submission of small workloads, and expected to be non-issue when the computation is performed on massive graphs.
Data Centers - Striving Within A Narrow Range - Research Report - MCG - May 2...pchutichetpong
M Capital Group (“MCG”) expects to see demand and the changing evolution of supply, facilitated through institutional investment rotation out of offices and into work from home (“WFH”), while the ever-expanding need for data storage as global internet usage expands, with experts predicting 5.3 billion users by 2023. These market factors will be underpinned by technological changes, such as progressing cloud services and edge sites, allowing the industry to see strong expected annual growth of 13% over the next 4 years.
Whilst competitive headwinds remain, represented through the recent second bankruptcy filing of Sungard, which blames “COVID-19 and other macroeconomic trends including delayed customer spending decisions, insourcing and reductions in IT spending, energy inflation and reduction in demand for certain services”, the industry has seen key adjustments, where MCG believes that engineering cost management and technological innovation will be paramount to success.
MCG reports that the more favorable market conditions expected over the next few years, helped by the winding down of pandemic restrictions and a hybrid working environment will be driving market momentum forward. The continuous injection of capital by alternative investment firms, as well as the growing infrastructural investment from cloud service providers and social media companies, whose revenues are expected to grow over 3.6x larger by value in 2026, will likely help propel center provision and innovation. These factors paint a promising picture for the industry players that offset rising input costs and adapt to new technologies.
According to M Capital Group: “Specifically, the long-term cost-saving opportunities available from the rise of remote managing will likely aid value growth for the industry. Through margin optimization and further availability of capital for reinvestment, strong players will maintain their competitive foothold, while weaker players exit the market to balance supply and demand.”
Data Centers - Striving Within A Narrow Range - Research Report - MCG - May 2...
B2 b fc credits performance deadlocks
1. B2B FC CREDITSPERFORMANCEDEADLOCKS
Jan 31, 2013 3:17 PM (
So thiswasabout gettingthe link"up"onboth ends,butwhataboutactuallysendingdataaccross those
links?Longerlinkscanactuallycontainmultiple fibrechannel frames!!The speedof lightisn'tinfinate,
so ittakestime (microsecondstosub milli seconds) foraframe to actuallytravel fromTx (transmitter)
to Rx (receiver).
If a fibre opticpathisan FCISL, theninknowingitsline bit-length,we cancalculate the maximum
numberof framesthat can simultaneouslyfit(source todestinationandback) onthat physical link.That
numberisthe exactanmountof buffertobufferscreditsyourISLswitchportswill require.
Why?
To safelystore-and-forwardframesatthe source end,while theyare still intransittothe destination
port,and acknowledgedashavingbeenreceived.Sothe sendingswitchwill keepthatframe ina buffer
while the ACKhasn'tbeenreceived.
To provide a“successfullysent”I/Oacknowledgmenttothe sendingportor HBA soit doesnothave to
waitfor the frame to reach the otherside of the link,andthe acknowledgmentthathasto come back.
So if youupgrade a port'sspeedfrom2 to 4 Gbps for example,the amountof framesthatcan be on the
linksimultaniouslydoubledandthe amountof B2B creditshasto be doubledaswell.
Please note thatalthoughthe linkbetweensitescanbe fastenoughandhave enoughbuffercreditsto
facilitate linerate communication,the receivingarraycan still be a bottleneckwhichcancause the
sendingside toslowdown.Considerforexample MirrorView orSRDF.The primarystorage array is
writtentoby a numberof hostsand the the back endof the primaryarray can handle the amountof
IOps.All I/Osthatwere writtenare alsosentto the secondaryarray to be replicated.If the receivingend
cannot handle thatamountof (write) IOpsbecause the numberof disksistoolow orthere'snot enough
cache available,the receiving(busy) arrayisprocessingthe incomingI/Osnotasfastas the ISLbetween
the sitescan provide andsoacknowledgementsaren'tsentbackto the primaryarray instantaniously,
but a little bitslower.Sointhe endthe sendingapplication receivesthisacknowledgementsomewhat
laterand we all knowthat comminucationisbuiltonsend/receive andACK,right?If the ACKcomesin
slow,the conversationisslowerthanyou'dlike ittobe.
2. So configuringthe linkbetweensitesisone thing,sizingthe recveivingarrayisanother(important)
thing!
Before we beginanexample of howtocalculate bufferrequirements,itisimportanttoknow the
numerical definitionof aFibre Channel Gigabit,aswell astounderstandthe structure of a Fibre Channel
Frame.
In the Fibre Channel world,one gigabitisdefinedtobe 1,062,500,000 bits(whichisnot
1024x1024x1024). Other Fibre Channel Gigabitvaluesare thenderivedfromthisreference definition.
For example,two(Fibre Channel) Gb=2 x 1,062,500,000 bits= 2,125,000,000 bits.To avoidconfusion
withthe traditional (nonFibre Channel) definitionof aGb, throughoutthisdocumentIwill use the
symbol Gbfcto mean“1,062,500,000 bits”,or 1 Fibre Channel Gb.
In summary:
1 Gbfc = 1,062,500,000
2 Gbfc = 2,125,000,000
4 Gbfc = 4,250,000,000
8 Gbfc = 8,500,000,000
10 Gbfc = 10,625,000,000
16 Gbfc = 17,000,000,000
Next,we showthe anatomyof a Fibre Channel Frame withnotes.
Start of Frame:4 bytesor 32 bits
StandardFrame Header:24 bytesor192 bits
Data (payload):[0– 2,112] bytesor [0 – 16,896] bits
CRC: 4 bytesor32 bits
End of Frame: 4 bytesor 32 bits
TOTAL (Nbrbits/frame):[36– 2,148] bytesor 288 – 17,184 bits
Notes:
The term byte usedhere means8 bits(notthe 10 bitsthat resultfrom 8/10 bitencoding).
The maximumFibre Channel frame sizeis2,148 bytes.
The final frame size mustbe a multiple of 4bytes.Thusthe Data (payload) segmentwill,asnecessary,
be paddedwith1 to 3 “fill-bytes”toachieve anoverall 4byte frame alignment.
3. The standard Frame Headersize is24 bytes.However,upto64 additional bytes(foratotal of an 88 byte
header) canbe includedforapplicationsthatneedextensive control information.Since the total frame
size cannotexceedthe maximumof 2,148 bytes,these additional Headerbyteswillsubtractfromthe
Data segmentsize byas muchas 64 bytes(perframe).Thisiswhythe maximumData(payload) size is
2,112 (because [2,112 – 64] = 2,048, whichisexactly2K-bytesof data).
The final frame,once constructed,ispassedthroughthe 8 byte to 10 byte conversionprocess.
In the FC world,1 Word = 4 x 8/10 bit encodedbytes(40bits).
Thenwe have the speedof light:
299,792.458 km/ s == 0.00000333564095 s / km, so that's 3.33564095 microsecondsperkilometer,
that's 3.3 millionthof asecondfor eachkm "traveled".
NOTE: thisisin vacuum!I'll adjustthe outcome later.The speedof lightinfiberisapproximately
200,000 kmper secondinsteadof almost300,000.
Nextwe take the lineardistance betweenthe twositesanddetermine:
How manysecondsittakesfor 1-bitto travel the one-waydistance linearbetweenthe twosites;thatis,
expressthe “distance”betweenthe twositesinseconds.Thisisdeterminedbythe speedof light.
Havingdetermined the one-waydistance insecondsbetweenthe twosites(afixednumber),we can
nowdetermine the maximumnumberof bitsthatcan exist,intransit,betweenthe twositesatanyone
time.Inotherwords,we can calculate the “distance”betweenthe twosites inbits(asopposedto
miles).Thisisdeterminedbythe speedof light,aswell asbythe rate at whichthe transmitting
equipment(e.g.afibre channel port) cancreate electrical variationsontothe medium(i.e.fibre optic
line).Inotherwords,howfastcan itpush bitsontoone endof the fibre opticline (usually1Gbfc,2
Gbfc, 4 Gbfc or 10 Gbfc).
Let say,for the purposesof discussion,thatwe have redundant(i.e.two) fibre opticpathsbetweena
primaryand secondarysite.
The lineardistance (asopposedtodisplacement)is91.732608 Km (or about57 miles) forone path,and
43.452288 km(or about27 miles) forthe otherpath.
Suppose the linkspeedis2Gbfc:
Distance (Km) Distance (secs) # of Bits(1Way) 8/10 FC bytes
91.732608 (57mi) 0.000305987044 650,222.468 65,022.2468
43.452288 (27mi) 0.000144941231 308,000.116 30,800.0116
Again:thisisin vacuuminsteadof fiber!Butstaywithme,I'll fix itlateron!!
4. Calculationsnotesforthe tablesabove:
Column2 indicates the amountof time (inseconds) ittakesone OPTICALVARIATION (howevermany
numberof bitsthat optical variationhappenstorepresent) totravel the one-waydistance specifiedin
columnone.Itcomesfrom multiplyingthe numberof secondsittakeslight totravel 1 km (i.e.
0.00000333564095 s / km),bythe numberof km traveledwhich,isspecifiedinthe firstColumn.
Column3 is the productof column2 (i.e.the line distance inseconds) andthe FCbitinsertionrate
(2,125,000,000). This columneffectivelyrepresentsthe amountof additional I/Othatcouldhave been
processedatthe host,had the response tothat I/Obeeninstantaneous;(actuallytwice thatamount,
since acknowledgmenttime forthose I/O’s,comingbackthe otherway,have to be accounted foras
well).
Column4 is Column3 dividedby10.This isdone to groupsingle bitsintransitintoequivalent8/10bit
“byte”quantities.The Fibre Channel protocol convertsevery8bitbyte intoa 10 bit equivalent(viathe
8/10 bit encodingalgorithm) before transmittingit.Sothe value inthisColumn4will determine the
numberof buffersneededforanISL switchport.
Frame Length8 (inkm) = (299,792.458 km/s) x (Seconds-Between- Inserted-Bitssecs/bit) x (Number-Of-
Bits-Per-Framebits)
Frame Length10 (inkm) = (299,792.458 km/s) x (Seconds-Between- Inserted-Bitssecs/bit) x (Number-Of-
Bits-Per-Framebits) x 10/8
Where:
Seconds-Between-Inserted-Bits=1/1,062,500,000 (for1Gbps FC) = 1/2,125,000,000 (for2Gbps FC)
Number-Of-Bits-Per-Frame=Variable dependingonDatapayloadand/orHeadersize.See table and
notesabove.Inmostcases the Headerwill be the Standardsize of 24 bytes.
Numberof bits/ frame
(after8/10 bitencoding)
Frame Length(km) @ 2 Gbps FC
(.000141078803764705 km/bit)
Numberof in-transitframes(1-
way) for 91.732608 km leg.
17,184 / 21,480 3.0303 km (2112 PL-bytes) 30.2170 frames(buffers)
Andagain: thisisvacuum,not infiber!!!Staywithme a little bitlonger....
Notes:
Column1 representsthe total numberof 8/10 bitsin a Fibre Channel frame (withastandardheadersize
of 24 bytes) atvaryingData payloads(PL).
5. Column2 representsthe productof the 10 bitvalue inColumn1, and (1/2,125,000,000). Thus,this
Column(Column2) essentiallyrepresentsthe lineardistance that1 (one) single frame consumesforthe
specifiedpayload(PL).
Column3 representsthe quotientderivedbydividingthe longest(worstcase) DWDMline distance (in
kilometers),bythe numberof kilometersperframe (calculatedinColumn2).Thus,thiscolumn
essentiallyindicateshowmanyadditionalONE-WAYframesworthof datacouldhave beenprocessedby
the host/application,hadthe response tothe firstframe beeninstantaneous.Inotherwords,thisishow
manyONE-WAY(notround trip) switch buffersyouwouldneedtoallow non-stoptransmission.Double
(i.e.roundtrip) the valuesinthiscolumn3to yieldthe numberof buffersrequiredof yourISLswitch
ports.
So inour example forrunning2longlinksusing2 Gbfc speed(invacuum):
the numberof B2B creditsneededforthe 91.7 km legwill be 2 x 30.2 = 60.4, soroundingupto 61 B2B
credits.
the numberof B2B creditsneededforthe 43.4 km legwill be 2 x (43.4/91.7) = 2 x 14.3 = 28.6, so
roundingupto 29 B2B credits.
Can't we make a rule of thumbout of this?
I admit,the numbersIusedare takenfroman example documentIhave lyingaroundsoI didn'tneedto
the math myself ,butmakinga ROT out of itbackfiresatme.
The easiestwayto calculate the rule of thumbisgettingbackto 1 km, soat 2 Gbfc speedthe numberof
B2B creditsneededfora1 km linkis2 x 30.2170 / 91.732608 = 0.6588 IN VACUUM.
Otherexampleswe cannowuse to more easilyrememberthese"nice"numbers:
2 Gbfc overa 2 km link= 2 x 0.6588 = ± 1.32
4 Gbfc overa 2 km link= 4 x 0.6588 = ± 2.64
4 Gbfc overa 20 km link= 40 x 0.6588 = ± 26.4
So whatwouldbe a nice numberto remember?4Gbfc requires1.3x the numberof km? 1.33 x the
numberof km? I guesswe have justcreateda rule of thumb!1.33 can be rememberedveryeasily!
Right?
So runninga 4Gb linkover44 km requires1.33 x 44 = 58.52 := 59 B2B credits.
6. A 2 Gbfc linkusingthe same distance (44km) will needhalf of that,so30 B2B credits.Sohalf the speed,
half the buffercredits,doublethe speed,doublethe buffercredits.
HOWEVER: the speedof lightinvacuumneedstobe devidedbythe refractionindexof fiber,whichis
about1.5.
BuffersRule of Thumb(takingthe actual speedof lightintoaccount):
4 Gbfc needs1.33 times1.5 timesthe distance inkilometer=1.995 buffersperkilometer
Thisextra1.5 is because of the refractionindex of the carrierused.Mostfibershave anindex of
approximately1.5,whichwill "slowdown"the speedof lighttoabout200,000 kilometerspersecond.
A longdistance connectionrunningat4 Gbfc needs2buffersperkilometer
A longdistance connectionrunningat4 Gbfc needs3.22 bufferspermile
(fromkm to mile youneedtomultiplythe outcome inkmby1.609, whichisnotthat easyto remember,
but youcan figure outa numberthatworksfor you,right?)
EMC as well asBrocade recommendaddinganextra20% B2B creditsto the amountyou justcalculated.
Message waseditedby:RRR Added20% extrasimilartothe bestppracticesbyEMC and Brocade.
ReportAbuse
Like Show2 Likes(2)
16. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
RRR Nov20, 2012 3:17 AM(inresponse toRRR)
If anyone thinksImade a mistake here,please informme,sowe can setthisstraight.
ReportAbuse
Like Show0 Likes(0)
17. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
Vincze Nov20, 2012 4:48 AM(inresponse toRRR)
7. Nice breakdown,RRR
ReportAbuse
Like Show1 Like (1)
18. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
RRR Nov20, 2012 6:01 AM(inresponse toVincze)
Thanks.AndI almostlostmy post!!WhenI was done Iaccidentlyswitchedtothe advanced editorand
nothinghappened....aaaargh!FortunatelyIcopiedthe html texttonotepadonlyaminute before,soI
was able torestore mypost. Tookme an hourand a half or so to type this,so I wasgladit wasn't lostin
the end.
So whatabout this1.33 factor for a 4Gb link?Doyouthinkit's correct?I can't findanyerrorsin my
calculations
ReportAbuse
Like Show0 Likes(0)
19. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
JonKlaus Nov20, 2012 7:32 AM(inresponse toMark)
AndI was afraidmy postwas long...yikes!
I originallycalculatedall buffercreditsbackfrom2gbit,whichis/was1creditper km.4Gbit wouldneed
2 creditsperkm,etc.
Don't forgetthat toomuch creditscan't harm performance (assumingyou're notrunningshortonother
links),andthatyoucan alwaysmonitorcreditusage onthe switch.CiscoDevice Managerhasa counter
that showsthe slackyou've got:
8. In thiscase it's an ISL that's maybe a couple of hundredmeterslong,soitcan cope with1 or 2 credits
dependingonactual speedanddistance.The portisneverthelessassigned32credits,of which32 are
still available(seebottom2numbers,CurrRxBbCreditsandCurrTxBbCredits.
ReportAbuse
Like Show1 Like (1)
20. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
RRR Nov22, 2012 1:43 AM(inresponse toMark)
Tonight(Europe time) we plannedtodothe TweetChat,butyesterday SeanThulin mentionedtousover
TwitterthatThursday Novthe 22nd isThanksgivinganditmightnotbe a goodday for the TweetChat.
So aftersome careful considerationswe decidedtomove the TweetChattonextMonday,November
the 26th. Same time,same place,newday!
ReportAbuse
Like Show0 Likes(0)
21. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
JonKlaus Nov23, 2012 4:26 AM(inresponse toRRR)
Happy Thanksgivingeveryone inthe USand aroundthe world!I hope the turkeytastedgood!
So..A lotof people are readingthistopic,butthe repliesare abiton the low side.Isthere no-one else
that wantsto share theirexperienceswithlongdistance links?Are longdistancelinksthateasy?
9. Please join the conversation;we won'tbite!
ReportAbuse
Like Show1 Like (1)
22. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
RRR Nov28, 2012 2:49 AM(inresponse toJonKlaus)
So we talkedaboutlight,multi mode andsingle modeaswell asattenuation.Sothe maximumdistance
isdefinedbyhowsensitive the Rx side is.
Alsowe explainedwhyBuffertoBufferCreditsare neededandhow youcalculate how manyof these
youneed.
So the nextthingwe'dlike todiscussisa wayto make optimal use of the expensivefiberlinkyoumight
have.A way to do thisisby usingcolors.Since lasersare monochromaticanywayusingasort of prismto
bundle colorscomingfromdifferentlasersintoasingle multicoloredbeamisthe solution!
TechniqueslikeCWDM,DWDM andEWDM were developedtofacilitateasmanycolorsas possible into
a single beam.
WDM standsfor wavelength-divisionmultiplexing.Multiplexingsaysitall:itcombinesmultiple source
intoone.
The bestknownWDM productis CWDM: Course WDM. In 2002 the standardwas setfor CWDM as using
wavelengthsbetween1270 and1610 nm as well ashaving20 nm spacingsbetweenthe "colors".The
mostcommonlyusedCWDMMultiplexers(the prisms) provide 8connectionsso8 sendinglongwave
ports can use the same physical darkfiber.Andalthougheachof these "colors"isnamedaftera visible
color(like brown,yellow,blue andsoon) the actual wavelengthsare farbeyondthe visibilityof the
humaneye.
The wavelengthsbetween1270 and 1470 are consideredunusablebecause of the increased
attenuation,soonly1470 to 1610 issupported.
10. The "colors"usedfor CWDM are:
1470 nm (gray)
1490 nm (violet)
1510 nm (blue)
1530 nm (green)
1550 nm (yellow)
1570 nm (orange)
1590 nm (red)
1610 nm (brown)
The mentioned colorsare the namesCiscocallsthese differentwavelength.Byusingcoloredtagson
each SFPthe humaneye can noweasilyidentifyeachwavelength.
Needlesslytosaythat to connecta networkonsite A usingforexample the 1530 nmwavelength,the
same wavelengthof 1530 nm is neededonsite B.
11. CWDM isunaware of the protocol usedtotransport data accross,so you can easilycombine tcp/ip,fibre
channel andevenanalogue tvsignalsaslongas eachsignal usesa differentwavelength.
ReportAbuse
Like Show1 Like (1)
23. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
JonKlaus Nov23, 2012 8:22 AM(inresponse toRRR)
So we knowhowto getthe signal acrosslongdistancesandwe know how to efficientlyloadthe link
usingbuffercredits.If we needmore bandwidththanone linkcandeliver,we cansenddifferent
wavelengthsoverone linkusingCWDMandDWDM to make thingsevenmore efficient.Butthere'sone
thingwe can't influence (much):latency.
Whenwe sendinformationbetweentwodistanceswe use light.The speedof lightis299,792,458m/s.
Thisis ina vacuum; infiberit'sa bit lower,somewherearound200,000,000m/s. Puttingitdifferently,
that's 200,000km/s or 200km/ms. This meansthatif we wantto cover200km of distance,we'll have
1ms of latencybetweenputtinglightonthe fiberandthe otherendseeingit.
To performreliable communications,we can'tsignal uni-directionallyandleave itatthat.We wantto
knowif the data arrivedat the other endintact.Thismeansthat we needtouse the bi-directional
latency,alsocalledround-triptime.Our200km linkisnow costingus 2ms of latency.Nottoo much
right?Well...helloSCSI!
FC data transportusesSCSIcommands.If we assume Iwill be the sendingstorage arrayonthe source
endand RRR will be the receiving storage onthe otherend,our(SCSI) conversationwouldbe something
alongthe linesof:
RTT direction1:HeyRRR,I wantto sendyou some data.
RTT direction2:Sure Jon,goahead.
RTT direction1:Okay,here isthe data.
RTT direction2:ThanksJon,gotit ingoodshape.
Or, since a picture saysmore than a thousandwords,a picture fromthe Brocade whitepapers:
12. Thismeansthat insteadof 1ms latency,our200km linkisactuallycostingus4ms of latency.Andthat's
onlyforthe linkitself.
If we send32KB of data,we also needtoput that onthe line.IgnoringFCframe overhead,at1Gbit/sa
32KB I/Otakes0,244ms to load.You can compare thisto a truck enteringthe highway:once the front
fendercrossesthe startingline,the backof itis still not"loaded"ontothe highway.
If we alsohave protocol convertersinbetweensource anddestination(because forexampleyouswitch
fromFC to FCIPand at the otherendback fromFCIPto FC) youwill needtoaddanother0,5ms for each
conversion.With4round-trips,you'll endupwith4msof protocol conversiontime.
So we've got4ms of delaydue tothe speedof light.We have 4ms due to protocol conversionsor
routers.Andwe have 0,244ms of latencydue tothe fact we can onlyloadthe data on the line at
1Gbit/s.Addit all up and at 200km youhave an additional latencyof 8,244ms.
Consideringthataunmirroredwrite I/Othatishittingwrite cache onlycostsus somethinginthe
neighborhoodof 0,5ms,thisisa lot.Switchon mirroringoverthatsecondarylinkandyourwrite latency
isnow 0,5 + 8,244 + 0,5ms = 9,244ms. If the secondarystorage systemisalsorunningoutof write cache
and needstoforce flush,you're ina worldof pain...
Moral of the story:be careful of synchronousreplicationoverlongdistances!
Andhave a goodweekend...
ReportAbuse
Like Show1 Like (1)
13. 24. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
JonKlaus Nov26, 2012 10:07 AM (inresponse toJonKlaus)
Andwe're off withthe tweetchat!Joinusontwitterwithhashtag#EMCATE or go the the following
website:http://tweetchat.com/room/EMCATE
Hope to see youthere for questions,commentsorjust a casual chat!
ReportAbuse
Like Show0 Likes(0)
25. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
RRR Nov26, 2012 12:13 PM (inresponse toJonKlaus)
We talkedaboutFCIP,buffervcredits,inorderdelivery,portgroups,sharedbandwidth,butalsoSONET
OC-48 to coveralmost1,000 miles.Prettyimpressive!
Lots of questions,lotsof answer,butstill alotof unansweredones.Needmore time.......
ReportAbuse
Like Show0 Likes(0)
26. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
RRR Nov27, 2012 3:50 AM(inresponse toRRR)
I fixedit.Itotallyforgotaboutthe speedof light,whichislowerinfibercomparedtovacuum.It'sabout
1.5 timesslower,sothe speedof lightinfiberisapproximately200kkm/s.
Thismeansmore frameswill live onthe link,somore buffersare needed.
ReportAbuse
Like Show0 Likes(0)
27. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
14. merlijnNov27,2012 4:59 AM(inresponse toJonKlaus)
Hi Jon,
Can the counteractuallydropto 0, or is1 the minimum?
Because there’salwaysatleast1 BBC for a port.
ReportAbuse
Like Show0 Likes(0)
28. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
JonKlaus Nov27, 2012 5:46 AM(inresponse tomerlijn)
Hi Merlijn,
Yes,the countercan actuallydropto 0. At that point,transmissionof new frameswillhaltuntill atleast
1 ACKisback and the counteris incrementedagain.
If you run the followingCLIcommandagainstan interface,forexample fc1/1:
sh intfc1/1 countersdetails|grepcredit
You will see somethinglike:
0 waitsdue to lackof transmitcredits
14576233333 transitionsof tx BB creditout of zero state
14 transitionsof rx BB creditto zero state
I had to google abit to make sense of it..apparently"transitionsof tx BBcreditout of zero state" does
not indicate aproblem.Ihave resetthe countersandit keepsincrementing;the othertwovalues
remain0. Thisis a 1km linkwhichhas32 credits,so ithardlyeverdipsunder31.
ReportAbuse
Like Show0 Likes(0)
29. Re:Ask The Expert:Discussingthe challengesof LongDistance Links
16. Partners
Company
My AccountLogIn
BROWSE PRODUCTS
Data Storage
Servers
ConvergedInfrastructure
Data Protection
Security
Networking
Dell ProductsforWork
SearchProductsby Name
ProductsA-Z
ShopEMC Products
ProductCommunity
Software Downloads
Contact Sales
Call 1-866-438-3622
Call 1-866-438-3622
Storage Categories
Data Storage OverviewEnterprise StorageEntry&Midrange StorageSoftware DefinedStorageAll-Flash
StorageCloudStorageStorageManagement&Networking
Storage ProductFamilies
ElasticCloudStorage (ECS)IsilonScaleIOUnityViPRVMAXAllFlashVNXXtremIO
ServerOverviewRackServersTowerServersModularInfrastructure ServersServersOptimizedforSDS
ConvergedInfrastructure Categories
ConvergedInfrastructure OverviewConvergedSystemsHyper-convergedSystemsHybridCloudPlatforms
17. Converged Infrastructure ProductFamilies
VxRackSDDCVxBlockVxRackFLEXVxRailVisionSoftwareVscale ArchitectureXCSeries
Data ProtectionCategories
Data ProtectionOverviewDataBackupandProtectionStorageDataBackupandProtectionSoftwareData
ProtectionSuitesCloudBackupandProtectionCopyDataManagement
Data ProtectionProductFamilies
Data DomainDataProtectionSuite
Detect,investigate,andrespondtoadvancedthreats.Confirmandmanage identities.Ultimately,
preventIPtheft,fraud,andcybercrime.
Explore productsandsolutionsfromRSA.
VisitRSA.com
NetworkingOverviewSwitchesRoutersandWirelessNetworking
OverviewDell LaptopsDellDesktopsDell ThinClientsandVDIProducts
Top of Form
Search
Bottomof Form
No resultsfound
No results found
BROWSE SOLUTIONS
All-Flash
Big Data
Cloud
ConvergedInfrastructure
Data Protection
ServerSolutions
HighPerformance Computing
NetworkingSolutions
Software Defined
Virtualization
18. IT Transformation
Modernize,automate andtransformyourbusiness.
Contact Us
Let’stalkabout yourIT transformation.
Contact Sales
Call 1-866-438-3622
Call 1-866-438-3622
SHOPNOW
Data Storage
Servers
ConvergedInfrastructure
Data Protection
Security
Dell ProductsforWork
Questions?
We can help.
Contact Sales
Call 1-866-438-3622
Call 1-866-438-3622
How to Buy
International Sales
Entry and Midrange StorageEnterpriseStorageAll-FlashStorageSoftware-definedStorageStorage
ManagementandNetworkingCloudShopAll DataStorage Products
Rack ServersServersOptimizedforSDSModularInfrastructure Servers
ConvergedSystemsHyper-convergedSystemsHybridCloudPlatformsShopAll ConvergedInfrastructure
Products
Backup andProtectionStorageBackupandProtectionSoftwareShopAll DataProtectionProducts
OverviewDell LaptopsDellDesktopsDell ThinClientsandVDIProducts
ShopAll SecurityProducts
19. DELL EMC SERVICES& EXPERTISE
Accelerate yourbusinessandITtransformationwithcloud,bigdata,andtechnologyconsultingand
services.
ProfessionalServices
CustomerService
Training& Certification
Dell IT Lifecycle Services
Contact Us
Let’stalkabout yourIT service needs.
InFocus
Gain insightsandexpertise onthe topicsshapingITTransformation—visitInFocus.
Contact EMC Support
Contact Sales
Call 1-866-438-3622
Call 1-866-438-3622
Top of Form
Search
Bottomof Form
EMC SUPPORT
MyService360
Supportby Product
Downloads
Community
Service Center
Learn aboutCustomerService
Create a Service Request
Manage Service Requests
Chat witha SupportAgent
ViewandManage CompanyInformation
21. Social at Dell EMC
RSS Subscriptions
Email General Inquiries
SalesInquiries
Call 1-866-438-3622
Call 1-866-438-3622
BusinessOverviewBrandInnovationatDell EMCProductSecurity
Top of Form
Search
Bottomof Form
No resultsfound
ViewAll SearchResults
CLOSE
MENU
Products
Solutions
Shop
Support