First published on TECHNET on Feb 25, 2016
System Center Data Protection Manager (DPM) 2012 R2 UR9
went live on 26
Jan, 2016. In the
, we focused on stability – in this release, we have added a slew of features to optimize on-prem and cloud backups for large Enterprise deployments. The feedback so far from customers has been very positive. Today, we are happy to announce that we are making it an “Important Update” on Microsoft Update, so it will be made available to more customers. As with all DPM URs, this update is cumulative and contains all the reliability fixes from the previous release. Below is the list of key improvements.
1. Never miss backup SLA for large file server
In some cases, for e.g. production server node shutdown due to power failure, DPM file tracking filter gets corrupted which marks replica as inconsistent. Previously, this required a consistency check to repair the corruption which could run into several hours or even days for large file server deployments. In this update we have made major improvements to optimize the filter repair process.
DPM now leverages USN journal (Update Sequence Number Journal) technology in Windows to track files that have changed since the last synchronization job, making the repair operation much faster. We have seen up to 60X improvements in the tests we have run with large file servers (see more details in the test table below).
To optimize it further, the new repair operation will be triggered automatically with-in 15 minutes of the filter corruption. So there is no manual intervention required. The repair operation will be displayed as a synchronization job in DPM and will also sync the replica to latest.
Table below summarizes the time to complete repair synchronization job based on our internal testing. We have focused on two common deployments, large number of small files and small number of large files.
# Files protected
# Files modified
Time for repair synchronization job
1 million files in a 1 TB volume, each file being a few MBs
1 MB (static file server with no changes)
20 GB (2% churn)
1 hour 15 minutes
80 million files in a 2 TB volume, each file being a few KBs
Please note, individual file server repair times will vary depending on number of files, size of each file and directory structure.
2. No production server restart required
Production Server reboot while upgrading to the new release of SCDPM has been a big concern with our customers. With this release, we have done an extensive review of all the cases which were causing a reboot and eliminated all cases, except filter driver update where it is unavoidable. All backup products that do incremental backups use a filter driver on the production server and need to re-boot whenever they make changes to the filter driver. Good news is we don’t anticipate making more changes in this part of the code so this should hopefully bring smiles to the customer and upgrading to latest URs is easier.
3. Reduced cache space for Online backups
In addition to backing up data to disk, DPM also provides an option to use
Azure cloud for long term retention and offsite storage
using Azure Backup. Previously, Azure Backup required cache space of 15% of the data source size for backup to Azure. This was an issue for customers with large data sources (for e.g. volumes greater than 10 TB). With the latest
Azure Backup agent
, we reduced the cache space requirements to less than 5% which is 3X improvement.
4. Increased retention for Online backups
DPM & Azure Backup has increased the number of recovery points for cloud backups. This enables flexible retention policies to meet stringent compliance requirements such as HIPAA for large enterprises. For those who are curious on what the new maximum number of recovery points is, it has been increased to 9999 from 366.
5. And More…
We have made a bunch of other optimizations to improve customer experience when using DPM. For more detailed list and to download the latest DPM please refer to DPM 2012 R2 UR9 KB article