Highlighted
8 Krypton

DD670 - need to know actual data amount going into nics

Hi,

We have a Data Domain DD670 running OS 5.1. We use ddboost with Symantec Netbackup 7.5 for backups on windows and unix servers.

I need to know how much data is actually entering the Data Domain nics over e.g. a weekly basis. What with all the deduplication at source going on it is hard to know whether reports generated refer to logical uncompressed data or actual bytes received. Netbackup reports only on logical backup size. I can get a report on individual backup sizes from the dd web gui that i think I understand (shows bytes, pre-lc, post-lc etc) but doesn't give me a daily or weekly data into the nic ingress amount.

The Reports section on the web gui includes the File System Daily Capacity Utilization report, a relevant section is below. I'm guessing this is:

Data Written (Pre-Comp) = actual raw data in gb received that may well represent data deduplicated at source

Used (Post-Comp)  = data size after further local compression

daily-capacity.JPG.jpg

Can anyone comfirm if Data Written (Pre-comp) represents the actual amount of data travelling across the lan? If not what is, and how do I find out the actual lan data figure?

Thanks

0 Kudos
5 Replies
8 Krypton

Re: DD670 - need to know actual data amount going into nics

Yes, pre-comp is pre-processed so this is your raw data coming to DD NIC (it might not be so raw in case of DD Boost, but still it is what is sent to DD appliance). ifconfig -a will give you traffic per interface so you can check that for overall traffic too.

0 Kudos
8 Krypton

Re: DD670 - need to know actual data amount going into nics

Thanks for your helpful reply! But the two sources of info do not seem to match. The above daily capacity report for one week gives a total of 54780gb, which is 53tb. But the ifconfig -a show my active lan cards have not received that much data. Uptime is 41 days  - I assume the RX receive stats are counting from the last restart. They report only

eth0a     Link encap:Ethernet  HWaddr 00:7b:FA:05:71:01 

          inet addr:94.207.64.76  Bcast:94.207.64.95  Mask:255.255.255.224

          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1

          RX packets:2832366416 errors:0 dropped:0 overruns frame:0

          TX packets:2673024645 errors:0 dropped:0 overruns carrier:0

          collisions:0 txqueuelen:100

          RX bytes:3068490745945 (2.7 TiB)  TX bytes:2591677861634 (2.3 TiB)

eth0b     Link encap:Ethernet  HWaddr 00:8C:FC:05:71:00 

          inet addr:94.207.71.198  Bcast:94.207.71.255  Mask:255.255.255.0

          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1

          RX packets:1865949944 errors:0 dropped:0 overruns:290 frame:0

          TX packets:739352230 errors:0 dropped:0 overruns:0 carrier:0

          collisions:0 txqueuelen:1000

          RX bytes:2652406323389 (2.4 TiB)  TX bytes:81671695467 (76.0 GiB)

eth4a     Link encap:Ethernet  HWaddr 00:1B:33:D7:F3:30 

          inet addr:94.207.170.63  Bcast:94.207.170.255  Mask:255.255.255.0

          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1

          RX packets:126481378 errors:0 dropped:0 overruns:0 frame:0

          TX packets:32179079 errors:0 dropped:0 overruns:0 carrier:0

          collisions:0 txqueuelen:1000

          RX bytes:161040626373 (149.9 GiB)  TX bytes:3396247475 (3.1 GiB)

eth4b     Link encap:Ethernet  HWaddr 00:1B:21:D7:F3:22 

          inet addr:94.207.170.68  Bcast:94.207.170.255  Mask:255.255.255.0

          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1

          RX packets:243056610 errors:0 dropped:0 overruns:0 frame:0

          TX packets:47745466 errors:0 dropped:0 overruns:0 carrier:0

          collisions:0 txqueuelen:1000

          RX bytes:325746922390 (303.3 GiB)  TX bytes:4829871269 (4.4 GiB)

This is a total recieve figure of only 5 - 6 tb.... but the daily capacity report weekly figure is ten times as much. Why are they so different - am I interpreting this wrong?

Thanks

0 Kudos
8 Krypton

Re: DD670 - need to know actual data amount going into nics

I can't say top of my head, but to draw good conclusion one would need to reset counters first and then compare numbers as they start from 0 and see where (and if) difference comes from.  For example, when you say daily capacity I'm not sure what exactly do you refer too (is it pre or post or is it cumulative post value - if last one it will always be bigger than actual traffic one of course).

0 Kudos
8 Krypton

Re: DD670 - need to know actual data amount going into nics

Hi,

The pre-comp figure is in the Daily capacity report image I posted in my first post. I understood that pre-comp is the data as it arrives (whether or not it is already de-duplicated at source), and you have confirmed that. But the pre-comp total from the above table gives a weekly total of 54780gb (approx 53tb) yet the nics RX receive size in the ifconfig -a only show that 5 -6tb have passed through combined nics, and that is after 41 days uptime. These two figures are very different. Perhaps the count on the nics resets daily or doesn't represent 41 days worth?

Is there a command to reset the nic count, or would it require a reboot?

0 Kudos
8 Krypton

Re: DD670 - need to know actual data amount going into nics

There are couple of ways (eg. http://www.thegeekscope.com/linux-ifconfig-how-to-reset-tx-rx-counters/), but it depends on the your access rights and also what you wish to achieve .  for example, I would never reboot box for this as I can easily read current stats and check them back in 24h.  You can script that too.  You can check traffic on switch and collect port stats so you do not get any DD involvement and so on...

0 Kudos