This topic provides a technical overview of EMC® Data Domain® Boost (DD Boost™) features. Configuration information for these features is described in Configuring the Data Domain system.
Distributed segment processing allows parts of the deduplication process to be performed by the EMC® Data Domain® Boost (DD Boost™) Library, which avoids sending duplicate data to the EMC® Data Domain® system that is configured as a vRanger repository. The option to distribute the processing is enabled or disabled on the Data Domain system; for more information, see Configuring distributed segment processing. The DD Boost Library negotiates with the Data Domain system for the current setting of the option and behaves appropriately. Therefore, the DD Boost Library offers two modes of operation for sending backup data to a Data Domain system: one with distributed segment processing enabled and the other with the distributed segment processing disabled.
• Potentially higher throughput, because the DD Boost Library sends only unique data to a Data Domain system versus sending all the data over the LAN. Throughput improvements depend on the redundant nature of the data being backed up, the overall workload on the backup server, and the capability of the backup server. In general, greater throughput is attained with higher redundancy, greater backup server workload, and backup server capability.
• For more information about local compression and its configuration, see the EMC Data Domain Operating System Administration Guide.You manage distributed segment processing by using the ddboost option commands, as described in Using the DD Boost commands.
In this mode, the EMC® Data Domain® Boost (DD Boost™) Library sends the data directly to the Data Domain system over the LAN. The Data Domain system then segments, deduplicates, and compresses the data before it is written to the disk. The DD Boost Library runs on the backup server.In this mode, the deduplication process is distributed between the EMC® Data Domain® Boost (DD Boost™) Library and the Data Domain system.
Parts of the deduplication process are run by the DD Boost Library so that only unique data is sent to data to a Data Domain system over the LAN. The DD Boost Library segments, computes IDs for the segments, checks with the Data Domain system for duplicate segments, compresses unique segments that are not found on the Data Domain system, and sends the compressed data to the Data Domain system. The Data Domain system then writes the unique data to disk.
vRanger supports a single host name and IP address per repository. For EMC® Data Domain® systems that require multiple 1 GbE links to obtain full system performance, you must set up multiple storage servers on the Data Domain systems — one per interface — and target the backup policies to different storage servers to spread the load on the interfaces.The Advanced Load Balancing and Link Failover feature allows for combining multiple Ethernet links into a group. Only one of the interfaces on the Data Domain system is registered with vRanger. The EMC® Data Domain® Boost (DD Boost™) Library negotiates with the Data Domain system on the interface registered with vRanger to obtain an interface to send the data. The load balancing provides higher physical throughput to the Data Domain system compared to configuring the interfaces into a virtual interface using Ethernet level aggregation.The Data Domain system load balances the connections coming in from multiple vRanger instances on all the interfaces in the group. The load balancing is transparent to vRanger. Because Advanced Load Balancing and Link Failover works at the DD Boost Library software layer, it is seamless to the underlying network connectivity and supports both physical and virtual interfaces.The data transfer is load balanced based on the number of connections outstanding on the interfaces. Only connections for backup and restore jobs are load balanced.
The Advanced Load Balancing and Link Failover feature can be used with other network layer aggregation and failover technologies. The Advanced Load Balancing and Link Failover functionality also works with other network layer functionality on the Data Domain systems, including VLAN tagging and IP aliasing. This functionality allows additional flexibility in segregating traffic into multiple virtual networks, which run over the same physical links on the Data Domain system.
• Since only one interface group is supported on the EMC® Data Domain® system, for connecting more than one backup server, a switch is needed in the middle.
File replication replicates a single file. The replication request is initiated by the application. By default, the backup servers set up replication jobs between two EMC® Data Domain® systems after authenticating them using the pre-configured EMC® Data Domain® Boost (DD Boost™) user name and password. If the replication feature is enabled, when an optimized job is set up by backup servers, the session between the source and destination Data Domain systems is encrypted using Secure Sockets Layer (SSL), ensuring all image data and metadata is sent encrypted over the WAN.Enabling this option on Data Domain system is transparent to the data protection application. When the data protection application requests Data Domain system to perform a replication job, the source and destination systems negotiate automatically to perform encryption without knowledge of the requesting data protection application. Replication uses the ADH-AES256-SHA cipher suite. There is no ability to configure a different suite.Replication is available to a Data Domain system with an installed Replicator license, and applies to all replication jobs on that system. Both the source and the destination Data Domain systems that are participating in replication jobs must have this option enabled for it to take effect.Replication can be used with the encryption of data-at-rest feature available on Data Domain operating system with the optional Encryption license. When replication is used with the encryption of data-at-rest feature, the encrypted backup image data is encrypted again using SSL for sending over WAN.
• Both the source and the destination EMC® Data Domain® systems must be running Data Domain OS 5.2 or later to use this feature.The low-bandwidth Replicator option reduces the WAN bandwidth utilization. This option is useful if replication is being performed over a low-bandwidth network (WAN) link. This option provides additional compression during data transfer and is recommended only replication jobs that occur over WAN links that have fewer than 6 Mb/s of available bandwidth.The low-bandwidth optimization option is available to EMC® Data Domain®s systems with an installed Replicator license. The option is enabled on a Data Domain system and applies to all replication jobs on that system.Enabling this option on Data Domain system is transparent to the data protection application. When the data protection application requests a Data Domain system to perform a replication job, the source and destination systems automatically perform the additional compression without involving the requesting data protection application.