Video Screencast Help
Give us your opinion and win with Symantec! Please help us by taking this survey to tell us about your experience with Symantec Connect, so that we can continue to grow and improve.  Take the survey.

not getting throuhput while running a client backup through local appliance

Created: 29 Jul 2013 • Updated: 29 Jan 2014 | 8 comments
nick17's picture
This issue has been solved. See solution.

Hello all,

I do have a window 2003 R2 server with service pack 2. earlier we were taking its backup through WAN and were getting superb throughput. but when we install a local appliance to the server and tried to take its backup it started with 30 KBPS. can someone help me on that

Operating Systems:

Comments 8 CommentsJump to latest comment

symantec040's picture
Earlier the backup was taken via WAN. Was it just regular backup or Client side Dedup Backup.
Which model of the appliance is used here, 5020/5220 ? 
In 5220, at boot grub level, by default the appliance is booted into WAN optimization mode. And if its a 5220, what capacity NIC's are used/bonded, 1Gbps/10Gbps.
What is the network connectivity throughput at client's end? 
is it first time backup you have tried with Appliance using client side dedupe? 
In either case, look for pd.conf file in /usr/openv/lib/ost-plugins/pd.conf, share the entries which contain "bandwidth" here.
And take help from Symantec Support to run SAS(/appcritical) test to verify the network readiness.
nick17's picture


Thanks for your reply...

here are the details:-

Earlier the backup was taken via WAN. It was Client side Dedup Backup

Its a 5220,  4 x 1 GB 802.3ad bonding

1000/Full on client side

on WAn we were getting throughput of 50-60 MBPS

entries of PD.conf from the local appliance which is also media server are:-

# Debug Log File Max Size (MB)
# Determines if server-side settings from storage unit should be used
# 0 - Server settings can override local pd.conf settings
# 1 - Local pd.conf settings will always be used; server settings are ignored
# Compression (0=Disabled, 1=Enabled)
# Encryption (0=Disabled, 1=Enabled)
# Optimized-duplication timeout (minutes)
# Optimized-duplication max bandwidth (KB/s, 0=No Limit)
# Optimized-duplication compression (0=Off, 1=On)
# Optimized-duplication encryption (0=Off, 1=On)
# Maximum amount of memory to use for fingerprint caching (MB)
# Maximum number of images to load for fingerprint caching (0-4096)
# Load fingerprint cache during incremental or synthetic backups (0=Off, 1=On)
# Load fingerprint cache if on local storage server (0=Off, 1=On)
# Load fingerprint cache from specified existing client/policy.
# Parameters are:
#   clienthostmachine is the exact host name shown within the backup policy
#   backuppolicy is exact backup policy name to be used for caching
#   date is the last day to use this configuration setting (mm/dd/yyyy)
# FP_CACHE_CLIENT_POLICY = clienthostmachine,backuppolicy,date
# Segment size for main data backup stream (KB)
# Size must be a multiple of 32 and fall in the range 32-16384
# Segment size for meta data streams (KB)
# Size must be a multiple of 32 and fall in the range 32-16384
# Don't segment files with these suffixes
# DONT_SEGMENT_TYPES = mpeg,jpg,mp3
# Preferred segment size in KB for different file types
# edb = Exchange, mdf,ldf = MS SQL snapshot, segsize64k = MS SQL stream
PREFERRED_EXT_SEGKSIZE = edb:32,mdf:64,ndf:64,segsize64k:64
# Max bandwidth backup/restore (KB/s, 0=No Limit)
# Max image fragment size allowed (MB)
# Specify the local NIC for backup/restore
# PDDO web service retry count
# PDDO web service call timeout (seconds)
# Agent side decryption and decompression on restore (0=off, 1=on)
# Buffer size to use when prefetching data for restore operations (B)
PREFETCH_SIZE = 33554432
# Time interval for retrieving stats from CR (seconds)
# By default 0 is used to disable caching and retrieve stats on demand.
# Longer intervals may help with larger concurrency.
# Enable the fiber channel for backup/restore. 0 is disable, 1 is enable.
# Client rebasing.  Parameters are:
#  clienthostmachine is the exact host name shown within the backup policy
#  backuppolicy is exact backup policy name used to protect the VMWare machine
#  date is the day a backup is to be performed (mm/dd/yyyy)
#CLIENT_POLICY_DATE = clienthostmachine,backuppolicy,date
symantec040's picture

ok. Thats strange!

is the less throughput observed each backup for the client? I ask you this - as I want to eliminate first time seeding issues.

And do you see similar issue with other client's side dedupe too ?

Are you able to capture the spoold log from the client. and media 5220 and check for gaps in the time delays if any ? 

Interesting thing here for me is - Are the initial speeds better than 30KBps ? or is just over time the speed graph starts to detoriate ?

nick17's picture

We have two clients at that location and earlier both were running over WAN going to master appliance in another state. Now we implemented local appliance, one client is giving marvellous throughput while other is giving only 30 KBPS. 

how to collect spoold logs? 

over WAN this client was giving good throughput and even now till the time issue got resolved we are running it over WAN. but on local appliance from very first day its throughput is 20-30 KBPS.

symantec040's picture


Spoold logs can be obtained here - <dedupe_partition>/log/spoold/spoold.log & <dedupe_partition>/log/spoold/storaged.log

What I also think here is the problem could be more at client side network settings! 

Because - Appliance won't act strange with specific client incase of client side dedupe. Because Appliance is on the receiving end.

Now, the other client which is giving good throughput - could you ask Network guys to compare the configurations on these two clients to the switch? And the connection between switch and appliance.

Look at the hops and ask the network team for a data flow path between the slow throughput client and your local appliance.

Also as the backup is active - on the local appliance,> Network> Statistics , to see appliance NIC activity

nick17's picture


here are the findings from network team...

please find mentioned below answers to your queries regarding file server and appliance network connectivity.

Network guys to compare the configurations on these two clients to the switch?

Both the file servers are connected to same switch same module and have same port configuration.

And the connection between switch and appliance.

Connection between switch and appliance is intact and working fine with 4 gig ports bundled in Po2.

Look at the hops and ask the network team for a data flow path between the slow throughput client and your local appliance.

There is no L3 hop as appliance and file servers are in same vlan but two L2 hops between appliance and file servers. The two L2 hops are connecting appliance and connecting file servers and both of these switches are directly connected with each other. The path is same for both the file servers communication with appliance.

Mark_Solutions's picture

As only one client is affect it is the client you need to look at mainly, though it could be a DNS issue on the appliance too

I would start by adding hosts file entries on the client and appliance so that they do not need to use DNS when talking to each other

Next check out the pd.conf on the client - it may even be worth deleting the new appliance server entry from its ost-plugins directory and allow it to be re-creatred in case it has anything not quite right about it.

Hope this helps

Authorised Symantec Consultant

Don't forget to "Mark as Solution" if someones advice has solved your issue - and please bring back the Thumbs Up!!.

nick17's picture

sorry guys, i forgot to put the solution of this query.

there was bandwidth throttling enabled for the client and entry was made in master server and that was provoking client to get the throughput.

We removed the entry and backup starts with good throughput.

Thanks all for your valuable suggestions.yes