10-14-2015 07:45 AM
Hi guys,
I've got an issue with Wanoptimization option. Although it's enabled, all the copies of my server in site B connected to Main site by 100Mb point to point circuit are using Non-Optimized traffic.
Furthermore, the copy (file server in vmware machine using another vmware machine as VM backup host in the same vmware host) has decreased the speed from 80 Mb/s to 50 Mb/s suddenly. I mean, previous weekends, the copy took 50 hours, now it's taking 77H copying the same data. I've just launched a test copying only O.S. disk and as you can see above it has a constant download speed of 50Mb/s
Other data: I've got set up resiliance network with between Master/Media server and file server, there is no "Accelerator" license (don't ask, long story...). File server is Windows 2012, Master and Media Server is Netbackup Appliance 5230, Vmware host ESX 5.1
So, I've got two problems, Wanoptimization seems not to be working, and the decreasing of speed.
Do you know where the problem could be?
Thanks in advance!
260 0.00 0 0 6.53 52 52210
270 0.00 0 0 6.63 53 53065
280 0.00 0 0 6.59 52 52700
290 0.00 0 0 6.73 53 53839
300 0.00 0 0 6.54 52 52351
310 0.00 0 0 6.28 50 50213
320 0.00 0 0 6.47 51 51773
330 0.00 0 0 6.36 50 50917
340 0.00 0 0 6.75 53 53986
350 0.00 0 0 6.70 53 53584
360 0.00 0 0 6.75 53 53974
370 0.00 0 0 6.75 54 54015
380 0.00 0 0 6.88 55 55013
390 0.00 0 0 6.48 51 51829
400 0.00 0 0 6.70 53 53604
410 0.00 0 0 6.74 53 53913
420 0.00 0 0 6.78 54 54223
430 0.00 0 0 6.73 53 53827
440 0.00 0 0 6.71 53 53678
450 0.00 0 0 6.83 54 54621
460 0.00 0 0 6.86 54 54874
470 0.00 0 0 6.41 51 51306
480 0.00 0 0 6.64 53 53086
490 0.00 0 0 6.74 53 53948
500 0.00 0 0 6.79 54 54327
510 0.00 0 0 6.72 53 53775
520 0.00 0 0 6.80 54 54406
530 0.00 0 0 6.90 55 55160
540 0.00 0 0 6.73 53 53826
550 0.00 0 0 6.47 51 51720
560 0.00 0 0 6.77 54 54162
570 0.00 0 0 6.85 54 54826
580 0.00 0 0 6.67 53 53377
590 0.00 0 0 6.02 48 48176
600 0.00 0 0 6.51 52 52061
610 0.00 0 1 6.39 51 51148
620 0.00 0 0 6.92 55 55386
630 0.00 0 5 6.68 53 53457
640 0.00 0 0 6.37 50 50982
650 0.00 0 0 5.46 43 43674
660 0.00 0 0 6.32 50 50530
670 0.00 0 0 5.91 47 47308
680 0.00 0 0 5.27 42 42180
690 0.00 0 0 5.31 42 42518
700 0.00 0 0 5.70 45 45625
710 0.00 0 0 5.52 44 44132
720 0.00 0 0 6.14 49 49126
730 0.00 0 0 5.04 40 40341
740 0.00 0 0 4.88 39 39028
750 0.00 0 0 5.67 45 45348
760 0.00 0 0 4.81 38 38453
770 0.00 0 0 4.49 35 35913
780 0.00 0 0 6.49 51 51912
790 0.00 0 0 5.50 44 44016
800 0.00 0 0 5.10 40 40794
810 0.00 0 0 5.20 41 41627
820 0.00 0 0 5.12 40 40992
830 0.00 0 0 5.19 41 41510
840 0.00 0 0 4.85 38 38791
850 0.00 0 0 5.00 40 40011
860 0.00 0 0 5.98 47 47817
870 0.00 0 0 6.17 49 49356
880 0.00 0 0 5.96 47 47714
890 0.00 0 1 5.61 44 44864
900 0.00 0 0 5.57 44 44533
910 0.00 0 0 6.18 49 49436
920 0.00 0 0 5.23 41 41806
930 0.00 0 0 4.95 39 39596
940 0.00 0 0 5.27 42 42135
950 0.00 0 0 5.80 46 46364
960 0.00 0 0 4.92 39 39388
970 0.00 0 0 5.81 46 46514
980 0.00 0 0 5.43 43 43431
990 0.00 0 0 5.14 41 41152
1000 0.00 0 0 5.85 46 46830
1010 0.00 0 0 5.97 47 47784
1020 0.00 0 0 5.99 47 47959
1030 0.00 0 0 5.54 44 44307
1040 0.00 0 0 5.40 43 43205
1050 0.00 0 0 5.67 45 45397
1060 0.00 0 0 5.65 45 45185
1070 0.00 0 0 6.24 49 49940
1080 0.00 0 0 5.69 45 45510
1090 0.00 0 0 5.80 46 46364
1100 0.00 0 0 5.96 47 47668
1110 0.00 0 0 5.14 41 41118
1120 0.00 0 0 5.56 44 44441
1130 0.00 0 0 6.40 51 51239
1140 0.00 0 0 6.78 54 54251
1150 0.00 0 0 6.64 53 53132
1160 0.00 0 0 6.79 54 54331
1170 0.00 0 0 6.34 50 50759
1180 0.00 0 0 6.95 55 55589
1190 0.00 0 0 7.01 56 56113
1200 0.00 0 0 6.87 54 54969
1210 0.00 0 0 6.92 55 55387
1220 0.00 0 0 6.79 54 54352
1230 0.00 0 0 6.81 54 54506
1240 0.00 0 0 6.99 55 55940
1250 0.00 0 0 6.89 55 55126
1260 0.00 0 0 6.82 54 54521
1270 0.00 0 0 6.90 55 55196
1280 0.00 0 0 6.98 55 55849
1290 0.00 0 0 6.99 55 55911
1300 0.00 0 0 7.04 56 56291
1310 0.00 0 0 7.00 56 56037
1320 0.00 0 0 6.99 55 55882
1330 0.00 0 0 7.03 56 56242
1340 0.00 0 0 6.99 55 55947
1350 0.00 0 0 6.92 55 55397
1360 0.00 0 0 7.02 56 56177
1370 0.00 0 0 6.94 55 55552
1380 0.00 0 0 7.04 56 56359
1390 0.00 0 0 6.93 55 55426
1400 0.00 0 0 6.93 55 55428
1410 0.00 0 0 6.97 55 55798
1420 0.00 0 0 6.98 55 55878
1430 0.00 0 0 7.01 56 56065
1440 0.00 0 0 6.92 55 55390
1450 0.00 0 0 6.89 55 55127
1460 0.00 0 0 6.94 55 55538
1470 0.00 0 0 6.94 55 55509
1480 0.00 0 0 6.97 55 55782
1490 0.00 0 0 6.89 55 55156
1500 0.00 0 0 6.85 54 54839
1510 0.00 0 1 6.84 54 54747
1520 0.00 0 0 6.74 53 53950
1530 0.00 0 0 6.81 54 54461
1540 0.00 0 0 6.76 54 54071
1550 0.00 0 0 6.86 54 54848
1560 0.00 0 0 6.83 54 54616
1570 0.00 0 0 6.86 54 54890
1580 0.00 0 0 6.83 54 54662
1590 0.00 0 0 6.88 55 55075
1600 0.00 0 0 6.83 54 54677
1610 0.00 0 0 6.77 54 54197
1620 0.00 0 0 6.59 52 52711
1630 0.00 0 0 6.38 51 51052
1640 0.00 0 0 6.57 52 52535
1650 0.00 0 0 6.72 53 53784
1660 0.00 0 0 6.79 54 54349
1670 0.00 0 0 6.79 54 54287
1680 0.00 0 0 6.83 54 54613
1690 0.00 0 0 6.75 54 54033
1700 0.00 0 0 6.70 53 53612
1710 0.00 0 0 6.88 55 55024
1720 0.00 0 0 6.81 54 54487
1730 0.00 0 0 6.75 53 53975
1740 0.00 0 0 6.69 53 53540
1750 0.00 0 0 6.81 54 54501
1760 0.00 0 0 6.35 50 50832
1770 0.00 0 0 6.68 53 53463
1780 0.00 0 0 6.85 54 54801
1790 0.00 0 0 6.82 54 54521
1800 0.00 0 0 6.37 50 50958
1810 0.00 0 0 6.78 54 54237
1820 0.00 0 0 6.82 54 54561
1830 0.00 0 0 6.70 53 53607
1840 0.00 0 0 6.79 54 54354
1850 0.00 0 0 6.72 53 53794
1860 0.00 0 0 6.66 53 53302
1870 0.00 0 0 6.53 52 52222
1880 0.00 0 0 6.54 52 52328
1890 0.00 0 0 6.85 54 54766
1900 0.00 0 0 6.48 51 51867
1910 0.00 0 0 6.77 54 54125
1920 0.00 0 0 6.76 54 54063
1930 0.00 0 0 6.66 53 53293
1940 0.00 0 0 6.90 55 55214
1950 0.00 0 0 5.18 41 41435
1960 0.00 0 0 4.61 36 36889
1970 0.00 0 0 4.35 34 34765
1980 0.00 0 0 4.30 34 34412
1990 0.00 0 0 4.26 34 34083
2000 0.00 0 0 5.91 47 47274
2010 0.00 0 0 6.83 54 54626
2020 0.00 0 0 6.42 51 51385
2030 0.00 0 0 6.48 51 51854
2040 0.00 0 0 6.38 51 51001
2050 0.00 0 0 6.88 55 55022
2060 0.00 0 0 6.84 54 54717
2070 0.00 0 0 6.51 52 52090
2080 0.00 0 0 6.98 55 55827
2090 0.00 0 0 6.97 55 55743
2100 0.00 0 0 6.97 55 55737
2110 0.00 0 0 6.97 55 55771
2120 0.00 0 0 6.58 52 52652
2130 0.00 0 0 7.02 56 56166
2140 0.00 0 0 7.00 55 55983
2150 0.00 0 0 6.87 54 54938
2160 0.00 0 0 7.00 55 55965
2170 0.00 0 0 7.02 56 56172
2180 0.00 0 0 6.99 55 55933
2190 0.00 0 0 6.57 52 52568
2200 0.00 0 0 6.72 53 53731
2210 0.00 0 0 6.63 53 53031
2220 0.00 0 0 6.77 54 54169
2230 0.00 0 0 6.71 53 53642
2240 0.00 0 0 6.41 51 51310
2250 0.00 0 0 6.42 51 51384
2260 0.00 0 0 6.91 55 55314
2270 0.00 0 0 6.94 55 55517
2280 0.00 0 0 6.92 55 55384
2290 0.00 0 0 6.77 54 54122
2300 0.00 0 0 6.84 54 54683
2310 0.00 0 0 6.96 55 55674
2320 0.00 0 0 6.81 54 54487
2330 0.00 0 0 6.57 52 52564
2340 0.00 0 0 6.49 51 51922
2350 0.00 0 0 6.62 52 52985
2360 0.00 0 0 6.80 54 54377
2370 0.00 0 0 6.73 53 53866
2380 0.00 0 0 6.84 54 54750
2390 0.00 0 0 6.50 51 51964
2400 0.00 0 0 6.76 54 54089
2410 0.00 0 0 6.81 54 54513
2420 0.00 0 0 6.44 51 51558
2430 0.00 0 5 6.78 54 54219
2440 0.00 0 0 6.82 54 54586
2450 0.00 0 0 6.87 54 54938
2460 0.00 0 0 6.72 53 53742
2470 0.00 0 0 6.82 54 54597
2480 0.00 0 0 6.76 54 54073
2490 0.00 0 0 6.26 50 50091
2500 0.00 0 0 6.70 53 53575
2510 0.00 0 0 6.70 53 53620
netbu.WANOptimization>
Solved! Go to Solution.
10-15-2015 06:46 AM
A nice overview of NetBackup Appliance WAN Optimization here:
https://www-secure.symantec.com/connect/articles/netbackup-appliances-network-optimization
...which describes how the WAN optimization feature may not result in tangible gains in all topology and connectivity scenarios.
10-15-2015 06:58 PM
10-14-2015 08:18 AM
10-14-2015 08:23 AM
Hiya,
Yep, I did, it's the output that I copied in the message. this one:
netbu.WANOptimization> Traffic 10 0
OPTIMIZED NON-OPTIMIZED
Time Offset MB/sec Mb/sec Kb/sec MB/sec Mb/sec Kb/sec
(sec) (avg) (avg) (avg) (avg) (avg) (avg)
----------- ---------- ---------- ---------- ---------- ---------- ----------
10 0.00 0 0 0.00 0 2
20 0.00 0 2 0.01 0 65
30 0.00 0 2 0.00 0 28
40 0.00 0 0 0.00 0 11
50 0.00 0 0 0.00 0 0
60 0.00 0 0 0.00 0 21
70 0.00 0 0 0.00 0 4
80 0.00 0 0 0.00 0 22
90 0.00 0 0 0.00 0 2
100 0.00 0 0 0.00 0 21
110 0.00 0 0 0.00 0 1
120 0.00 0 0 0.00 0 3
130 0.00 0 0 2.02 16 16130
140 0.00 0 0 3.26 26 26097
150 0.00 0 0 2.43 19 19402
160 0.00 0 0 3.22 25 25770
170 0.00 0 0 2.17 17 17345
180 0.00 0 0 3.52 28 28181
190 0.00 0 0 6.60 52 52765
200 0.00 0 0 6.81 54 54462
210 0.00 0 0 6.85 54 54787
220 0.00 0 0 6.39 51 51097
230 0.00 0 0 5.89 47 47152
240 0.00 0 0 6.63 53 53062
250 0.00 0 0 6.62 52 52958
260 0.00 0 0 6.53 52 52210
270 0.00 0 0 6.63 53 53065
280 0.00 0 0 6.59 52 52700
290 0.00 0 0 6.73 53 53839
300 0.00 0 0 6.54 52 52351
310 0.00 0 0 6.28 50 50213
320 0.00 0 0 6.47 51 51773
330 0.00 0 0 6.36 50 50917
340 0.00 0 0 6.75 53 53986
350 0.00 0 0 6.70 53 53584
360 0.00 0 0 6.75 53 53974
370 0.00 0 0 6.75 54 54015
380 0.00 0 0 6.88 55 55013
390 0.00 0 0 6.48 51 51829
400 0.00 0 0 6.70 53 53604
410 0.00 0 0 6.74 53 53913
420 0.00 0 0 6.78 54 54223
10-14-2015 08:30 AM
10-14-2015 08:42 AM
10-14-2015 03:57 PM
The output is when this copy (and only this copy) is working. I've only got one appliance in Main Site.
About latency, sorry, it's in Spanish but I think it's understandable:
Estadísticas de ping para 10.81.50.22:
Paquetes: enviados = 3626, recibidos = 3626, perdidos = 0
(0% perdidos),
Tiempos aproximados de ida y vuelta en milisegundos:
Mínimo = 9ms, Máximo = 26ms, Media = 9ms
So, no packet losts, and Latency Average = 9ms
10-15-2015 06:00 AM
10-15-2015 06:46 AM
A nice overview of NetBackup Appliance WAN Optimization here:
https://www-secure.symantec.com/connect/articles/netbackup-appliances-network-optimization
...which describes how the WAN optimization feature may not result in tangible gains in all topology and connectivity scenarios.
10-15-2015 08:29 AM
So, I shouldn't be worried about optimization network as my latency is good, right?
About the speed problems, right, there wasn't any change. This is the complete copy. It's done in weekend when nobody is in the office so the circuit is completely free.
10-15-2015 06:58 PM
10-20-2015 02:35 AM
Hiya,
I did another test this weekend. Instead of use vmware policy I changed it to Windows one. As there is two data volumes (E & F) I'm using multiple data streams. Copying from both units at the same time Netbackup is using more than 8600 Kb/s instead of 5400 as it does with vmware policy. But, none of the copies never goes faster than 4300Kb/s although one of them has finished.
To be honest, I was investigating about your comment "In order to fill 100mbps pipe at 9ms I would have to put about 110kB of unacknowledged data on the wire" and I don't know how to do that.
10-20-2015 03:59 AM
Although this is Windows article, it is anintroduction to one facet of RSS (Receive Side Scaling):
https://msdn.microsoft.com/en-us/library/windows/hardware/ff556942(v=vs.85).aspx
There's a lot of topic/discussion on the internet about the trade-off / balancing between packet size, bandwidth, latency, buffer-size. All these technologies are converging in an attempt to fully utilize available network bandwidth in the most efficient manner - essentially covered by RFC1323:
https://en.wikipedia.org/wiki/TCP_window_scale_option
.
Most of this is supposed to be hidden by the OS (at both ends), the NICs (at both ends), the NIC drivers (at both ends), and of course the switching infrastructure in between. So, there's no NetBackup setting for any of this (apart perhaps from client side buffer size):
https://www.veritas.com/support/en_US/article.TECH28339
.
What WVT is doing is showing you the fundamental breakdown of latency, wire speed and how much unacknowledged data one can get in flight. Basicaly, WVT knows this so well that he is able to roughly calculate (in his head) the fundamental figures.
.
In your figures you say Kb/s.. do you really mean kilo-bits per second? Or did you mean to say KB/s kilo-bytes per second (i.e. what NetBackup shows). Remember, network guys/girls talk in bits, whereas storage and backup guys/girls talk in bytes.
10-20-2015 06:20 AM
So now we have more bandwidth being used because we have two flows going instead of one. But when one stream runs, the TCP window isnt scaling up to use more of the pipe. It doesnt mean anything is broken. Read up on Bandwidth Delay Product on interwebs.
10-27-2015 08:06 AM
Hi guys,
Sorry, I was talking in KBytes :)
So, as storage and backup guy it's a little difficult to me understand the doc you sent me.
BTW, My thought is why previous weeks Netbackup was using the full circuit (near to 100 Mb/s, bits right now) and since 3 weeks ago, it's copying on half of speed. The only thing that I know is I don't set up anything new in windows servers or netbackup or circuit. If I copy a 4GB files from my computer to the copied server I get almost 100Mb/s. I guess it's not the same copy from my windows machine to another windows machine than netbackup copies.
10-27-2015 08:56 AM
Has a network admin implemented any traffic shaping? Maybe perhaps QoS for specific protocols/addresses/ports/timeframes?
What speed do you get if you copy the same 4GB file from the replication source server to the replication target server?
10-28-2015 08:41 AM
Sdo, our network admin implemented a QoS for Call Manager (VoIP). I asked for stopped it and, theoretically, it was stopped. Do you know any case where QoS for VoIP interfered with Netbackup?
About speed, just a remind:
- Netbackup Appliance and my computer (windows 7) are in Main Site
- Server Copied (windows 2012) in Secondary Site
So, copying from server copied to netbackup share folder it takes 850sec., 3638 KB/s. I was looking to the graphics and the speed wasn't stable. I mean, it reached near to 10MB/s and decrease to 11KB, then increase to near of 10 MB again.
Coping from server copied to my computer share folder it takes 335sec., 9231 KB/s. The connection was stable all the time, near to 10MB
10-28-2015 09:16 AM
Not heard of any direct interference. What I was concerned about was whether QoS was implemented in brutal manner affecting all point-to-point conversations. However, if QoS was implemented for specific point-to-point address which are nothing to do with NetBackup then you should not have been affected, unless the CPU or buffering overhead required for the network WAN entry/exit points is so intense that it slows down all traffic, maybe? Maybe when QoS is enabled on the devices that it was enabled on - maybe this means that all traffic has to be buffered through a second stage so that it can be inspected to see if any QoS rules apply, maybe? Do you see all these 'maybes'? These definitely mean that we do not understand what the QoS implementation is actually doing. A known unknown, so to speak. So, we're slightly in the dark right now, and we can only fully understand the problem if we fully understand all of the devices, all of the stages and and all of the network hops. :) Do you see how your network admin feels that he/she can confidently say that you won't have been affected. But how can they/you prove that?
Apologies Angel, but it's still not clear to me that your copy test was actually between the two devices which suffered the original problem? Can you re-state which two devices suffered inter-device comms throughput issues, and confirm that your test was between these two very same devices and in the same direction that was originally having a problem? Maybe create a table/list of tests, clearing identifying source and target and copy method/protocol and observed average speed, and whether QoS was enabled for VoIP, e.g. maybe six columns of:
DateTime, Source, Target, Protocol, Avg.KB/s, QoS Y/N
.
Also, I wouldn't use Windows File Explorer for my copy testing, because Windows will try to do all sorts of buffering - and so what you see from a user perspective is a saw tooth performance graph visual artifact because Windows will itself read a large chunk of the file and empty its own buffer next. So, instead try a speed test using FTP or SFTP or SCP.
10-28-2015 09:21 AM
It wont matter if QOS for VOIP is affecting anything because if you turn it off phone calls will sound like Mcdonalds drive thru speaker.