Dedup Issues Netbackup 7.5.0.1

[Spartacus81]

Friends,

I have created brand new Netbackup 7.5.0.1 enviornment (1xMaster 2xMedia 7.5.0.1 - all virtual). I have seeded 6TB worth of data shipped across to data center on a USB from different remote sites and back it up using a local client (Seed123). Each media server is responsible for three sites hence 3 TB worth of data seeded to Media server 1 and rest to 2. Please note i used Media server deduplication at the time of seeding. 

now i have all these remote sites configured with Netbackup and course i expect the next backup to be quicker. I made the change in pd.conf file on the client so that the fingerprints from the seeding client can be loaded but its taking ages to run..

FP_CACHE_CLIENT_POLICY = Seed123,gssuanbmed02_Standard,07/04/2012

does anyone have any idea how much a subsequent backup of 1.7TB of data followed a seed will take to run over a 10meg line?

also, does anyone know any other better way of carrying out the seeding process? 

[Marianne]

Seems the FP_CACHE_CLIENT_POLICY entry in pd.conf on the client is the only step that is needed for 7.5 - according to this TN: http://www.symantec.com/docs/TECH144437

No personal experience - hope other Connect members will share real life experience.

[Spartacus81 · Accepted solution]

think i got the bottom of this...

It worked for me as i can see the cache being loaded in nbostpxy logs..

ache_util_load_fp_cache_nbu: most recent backup client=spartacus, date=1341343115, copy=1, frag=1, ext=.fmk

20:55:20.050[2468.2520][DEBUG][dummy][60:bptm:3945:gannicus][DEBUG] PDSTS: cache_util_load_fp_cache_nbu: threshold=1341343099

20:55:20.051[2468.2520][DEBUG][dummy][60:bptm:3945:gannicus][DEBUG] PDSTS: cache_util_load_fp_cache_nbu: adding 'spartacus_1341343115_C1_F1.img' to cache list (1)

20:55:20.051[2468.2520][DEBUG][dummy][60:bptm:3945:gannicus][DEBUG] PDSTS: cache_util_load_fp_cache_nbu: done readdir

20:55:20.051[2468.2520][DEBUG][dummy][60:bptm:3945:gannicus][DEBUG] PDSTS: cache_util_load_fp_cache_nbu: readdir completed

think with dedup we just need to patient.. my experience is, once the seed is done than sebsequent backup followed the seed on the live client will take long to complete as it synchronize the seeded fingerprints with the actual client. that process can be time consuming depends on the amount of data.. once the first backup has completed the 2nd backup will be lot smaller and way too quicker.. example:

Data seeded from client1 at data center sitting next to media server: 1.7TB took 40hrs.

First backup over a 1GIG link on the live server to that media server took nearly 25hrs (i would suggest update the FP_CACHE_CLIENT_POLICY on the client with the seeding client if you dont it will still work)

Second Backup: less than 20mins.. job stats: only scanned the changed data and CR sent was the changed blocks..