Netbackup 6.5 first observations!

[AKopel]

We installed Netbackup 6.5 last week and things are looking pretty good so far.
We upgraded from Netbackup 6.0 MP4. Upgraded our Master and 10 Media Servers. So far we have left (most) of the clients alone.

High level great things:

- Unlimited tape sharing in SSO - any media server can read AND WRITE any tape that is configured in the same Server Group (You can now GROUP sets of media servers)
- Support for Exchange 2007 Streaming Backups (this was our big rush in installing 6.5)
- Baremetal restore INCLUDED in the product (not an extra SKU now!!!)
- Client Encryption INCLUDED in the product (same thing!)
- Ability to push MP and patches via LiveUpdate (fully controllable by NBU administrator)
- ONE SKU for all Database Agents (SQL,Sharepoint,Exchange) and they are fully interchangeable. IE, if you 'currently' own 10 Exchange Agents, 5 Sharepoint, and 12 SQL, you now just suddenly own a POOL of 27 DATABASE agents. This is big for us as we are REDUCING exchange servers with Exchange 2007 but INCREASING sharepoint servers like candy lately.


Other non-published 'goodies' I have noticed:

- In the windows GUI Activity Monitor, CURRENT FILE: now works and shows you the currrent file it is backing up (this never worked for us before)
- Data Classification: (can label data different ranks (i.e. Bronze, Silver, Gold etc) and apply then towards different Storage Lifecycle Policies. (I haven't played with this much yet, but it looks promising)
- Can freeze, unfreeze, suspend, and unsuspend media from the gui now.
- Reporting about DISK is much improved. Separate reporting sections on "Images on Disk" "Disk Logs", and "Disk Storage Unit Status"
- Inactive Policies actually show a different icon in the GUI so you can see them at a glance (instead of having to go to "Summary of all Policies" and sort by "Active"!!!)
- New Storage Unit Group Selection Criteria: (Prioritized, and Load Balance): So in addition to Round Robin and Failover (which were in 6.0) Prioritized is Choose the first SU in the list that is not busy , down, or out of media. Load Balance is a capacity-managed approch that takes into account the following (CPU/memory Utilization, Number of NBU jobs running on media server, amount of disk space available to media server)

Some 'not-so-fun' observations:

- When installing the Windows Client, there is a new service called "Netbackup SAN Client Fibre Transport Service" This is for the new SSO for disk option. If you don't have it licensed, it does not start (and it is set to Automatic start by default). We ended up scripting the client install to set it to MANUAL by default.

- Some custom column layouts were wiped out (i.e. my Volume Pool Layouts) This was a pain for our operations folks as we have them eject tapes on Mondays by "Description" which we populate for them. Their Description field was gone so we had to show them how to readd it.

Overall, it seems like a pretty solid upgrade (ESPECIALLY compared to the 5.x - 6.0 (which we painfully went through) Supposedly, the core of this is built off of 6.0 MP4 so I wasn't as worried about waiting for the first MP.

I still have yet to look to see if there are any improvements to NOM in 6.5.. Maybe in the next few weeks.


AK

[Stumpr2]

Thanks Aaron!

I hope others will follow up with their experience with the upgrade in this thread.

[GoPhins_]

How are your Outlook 2007 backups performing? (I understand that not all functions are available in this release - not till 6.5.1)

Have you had to restore anything to outlook 2007?

Thanks, we are debating upgrade... hear that 6.0MP5 + a patch will handle outlook 2007 backup and restore?

[AKopel]

I assume you are talking about EXCHANGE 2007 backup/restore? Then yes, it handles it just fine. They do not yet support VSS backups for Exchange 2007 (yes, I believe in 6.5.1 it will be supported) but it is working just fine using the 'legacy' streaming backups. The only drawback, is that legacy backups have to be done off of the Active Node in a CCR cluster where with VSS, Microsoft supports doing backups off the Passive node. Not the end of the world for us at this point as we can now just do Weekly fulls on the weekends and do Cumulative Incrementals of the logs during the week. Plus, one drawback to VSS backups (which we were playing with Backup Exec 11d before NBU 6.5 came out) is that they have to be done sequentially. We have LTO2 so we peaked out at around 30-40MB/sec. With the legacy streaming backups, we can backup multiple stores at a time, so we can saturate our Gig NIC with 2-3 streams. We have a dedicated backup NIC that is clustered and fails over with the Exchange Nodes so all backup traffic goes over that NIC anyways.

Alternate server RSG restores work as advertised like this too. So we don't have to dedicate space for Recovery Storage group on our big CCR Clusters.

Aaron

[Chris_Babcock]

Thank You for your time and effort putting this out here for us!

I am currently looking at going to 6.5.

Chris

[Wes_T__Gator]

Is it possible to go from 5.1 to 6.5 or will I need to install 6.0 first?

[AKopel]

Here is the 5.x to 6.5 upgrade guide:

http://seer.entsupport.symantec.com/docs/290141.htm

Aaron

[duucfho]

We have also upgraded from 6.0 to 6.5 on a Windows 2003 master.  Will upgrade the media servers (win and AIX) within a few weeks.  So far, everything has gone smoothly. The new GUI has some nice changes, though I haven't had a chance to really get into the changes.

Exchange 2007 has backed up successfully so far, and I am currently testing backing up VMWare ESX3.  However, can't really get anywhere until I get the media server up to 6.5

[Patrick_Whelan_]

How did you get Exchange 2007 to work? I was at a site last Friday and we could not get it to work. Called support and they said it wasn't supported until either 6.0MP6 or 6.5.1. And they implied that MP6 would be available before 6.5.1.

[duucfho]

Our exchange 2007 (which is a test environment) is a 2-node CCR cluster, and I am backing it up through a backup NIC using the exchange cluster resource name.  I am just doing a full backup up of the live resource, as that is the only thing that is really supported right now.  Something is better than nothing...

Here is some info I got from Symantec.  Once 6.5.1 and

My original plan was to backup the offline node, but I don't believe that is supported until the next update.  Also, I have not yet attempted to do a snapshot backup

[duucfho]

We have a 2-node CCR exchange 2007 cluster.  I am currently only doing a full backup of the database using the exchange cluster resource name (over a backup network).  That's really the only thing thats supported right now.

According to some info I got from a Symantec rep, I will have to wait until 6.5.1 and 6.5.2 in order to do the really good stuff: offline node backups, snapshots, mailbox level backups, etc.

Info from my Symantec rep:

NetBackup 6.5
In August 2007, NetBackup 6.5 GA supports:
• Database backup & recovery
• Microsoft Cluster Server environments
In 6.5.1, planned for DecQtr07, support will be added for:
• Mailbox backup & recovery
• Veritas Cluster Server environments
• Local and off-host snapshot backups
• VSS snapshot providers: Windows COW, VxVM, and HW Arrays (TBD)
The objective is to support the current NetBackup for Exchange capabilities with Exchange 2007.
In 6.5.2, planned for MarQtr08, support will be added for:
• NEW – Instant Recovery from snapshots for Exchange 2007 and Exchange 2003
In 6.5.3, planned for JunQtr08, support will be added for:
• NEW – Mailbox restore from a database backup
• NEW – Exchange 2007 on Windows Server 2008
• NEW – Backup & recovery of replicated Exchange databases
(via native Exchange Local or Clustered Continuous Replication

[duucfho]

Sorry, didnt mean to post a reply twice.

I'm new to the internets.

[Stumpr2]

That's quite allright. No problem.

Thanks for participating.

[StorageGuy201]

Thanks for the info guys. I'm curious about streaming backups. How are the log files managed? Do the log files get truncated and if so is the log shipping to the passive node affected (for e.g. premature truncation of the log files)?

What is the advantage of doing the streaming backup using NetBackup vs. NTBackup in this case?

[Ron_Cohn]

Any observations or comments on Vault?

[Rakesh_Khandelw]

 6.5 is a great, stable product with lots of additional features.  Cople of things more like day-today activity related I would like to share -

- By default there is IP caching within NB java console, which means if you go to Host properties and click on clients, it will take longer time to show the client list (because it is doing a IP address resolution and caching in background). To change the default behaviour a change must be made to the nbj.conf file located in /usr/openv/java.

This is the parameter:

Set FORCE_IPADDR_LOOKUP=0 in nbj.conf. Caching of InetAddresses is now expected behaviour as of NB 6.5 if FORCE_IPADDR_LOOKUP=1.

- People who are upgrading their current enviornment to 6.5, be careful when you push the client software for UNIX/LINUX from master server, you may have to change your Hardware Type and OS to new value (in cases where you were using a different OS type in previous version of NBU because your hardware type/OS was not listed in client selection).

- There shouldn't be any file with numbered name, e.g. 0 or 1 or 2 or...........  in <install_dir>/netbackup/db/class directory. This will cause image DB cleanup process to hang.

- If you have "override media owner" selected in your catalog backup schedule, your catalog backup may hang for hours.

[GoPhins_]

Rakesh,

concerning:

'override media owner'  do you know why it would make the catalog backup run longer? (or hang?)

How much quicker did the catalog backup run after the change?

[Rakesh_Khandelw]

 What I have shared is linited to my personal experience with NBU 6.5, it does not include any field data/feedback.  

I do not have any answer at this time on how it makes the catalog backup slow but the difference was huge, couple of days Vs less than an hour.

[Jim_Miller]

I upgraded my 5.1 Windows environment about a month ago. I have a master/media and 1 additional media server. I followed the 5.x upgrade guide and all went well.

I only ran in to 1 major issue. The additional media server is a large file server that contains 3 TB of files. This server is 2003 R2 and we are using DFSR to replicate the server to a remote site for DR purposes. 6.5 has a different behaviour and tries to back up any DFSR content under the shadow copy components. I have a case open with engineering to get this corrected. 

The change seems very similar to what users of Backup Exec 11D are reporting in the link below

https://forums.symantec.com/syment/board/message?board.id=103&message.id=56998

[Ron_Cohn]

Does this behaviour exist in NetBackup 6.0 MP5?