When thinking about data protection of business-critical resources, we often think about certain components that make backups and replication possible. Often, the focus is on the physical server infrastructure with a focus on the disk subsystem. However, there are other crucial factors at play when it comes to having a successful, effective, and efficient data protection infrastructure. One of the critical infrastructure components that often gets overlooked with data protection infrastructure build out is the network. How important is the network when it comes to data protection including backups, replication, backup copies, etc? What are network considerations when designing backup and replication strategies?
How Important is the network in backup design?
The network is extremely important in all aspects of IT infrastructure. After all, it is the plumbing that connections endpoints together and allows packets to get from one place to another. Network design and considerations should always weigh into any effective and efficient data protection design in an organization. Networks are an important consideration for backups for the following reasons:
- Generally, organizations do not have a separate “backup” network as this is cost prohibitive. So typically, backups traverse the same network as production network workloads
- Replication and Backup Copy traffic generally traverses WAN connections. This can create bandwidth contention between production and replication traffic, which must be considered
- Network Addressing has to be considered in/during a true DR failover scenario
- Production IP addresses may change in the replicated DR facility – routing, layer 2, client connectivity, and DNS have to be considered
Let’s take a look at these concerns one by one and how we can take this into consideration with designing data protection jobs and infrastructure. Also, what are backup technologies with modern backup solutions that can assist with some of the network challenges faced with organization’s data protection needs.
Backup Traffic on Production Networks
Generally, organizations do not have a separate backup network when it comes to carrying the backup traffic from servers to backup repositories. Some organization may use separate VLANs for some servers but the added complexity and configuration per server and all the physical network infrastructure can be costly in terms of time and management. In today’s high bandwidth networks in most enterprise datacenters, typically there may be enough headroom in production networks to absorb the extra network traffic during backup windows.
Modern backup solutions such as Vembu BDR Suite include technology to offset load on production networks. Vembu VMBackup interfacing with VMware vSphere, selects data transport modes based on VMware backup configuration. Vembu is able to utilize a mode call HotAdd Transport Mode that provides a LAN-free data transfer. It works by allowing virtual disks to be accessed as local disks by a backup proxy. The proxy then reads the locally attached disks and transfers backups to the Vembu BDR Server.
Additionally, SAN Transport mode is used predominantly when the backup proxy runs on a physical machine and the virtual disks are stored on an iSCSI or Fibre channel SAN.
Using these modern techniques to be able to alleviate network bandwidth contention can prove to be very beneficial, especially in environments with limited bandwidth. Organizations may need to schedule and align backup jobs as well to alleviate contention with production workload traffic.
Using backup solutions that utilize changed block tracking as well for virtual environment workloads can reap massive savings in data copied as well. Only changes made in a virtual machine since the last backup are copied as an incremental restore point. Thus, data that is copied is minimal compared to backup solutions that may copy everything on each backup iteration.
Replication and Backup Copy Traffic and WAN bandwidth Contention
Along the lines of bandwidth contention on production internal networks, bandwidth contention comes into play when thinking about replication traffic to DR facilities or backup copy operations to a secondary location. Business WAN connections are generally expensive, especially for high bandwidth connections. When thinking about replicating virtual machines to a DR facility or backup data to a secondary location, organizations will want to utilize technologies that make the replication process as efficient as possible.
Again, changed block tracking or CBT comes into play in a major way when thinking about only replicating changes across a WAN. This can conserve massive amounts of bandwidth. The replicated virtual machine is updated with the changed blocks since the last replication iteration. Organizations will most definitely want to make sure replication jobs are utilizing changed block tracking.
One of the initial problems when thinking about bandwidth and replicating or copying backup data to a DR facility is “seeding” those “full” backups across to the DR facility. The initial transfer can’t benefit from the changed block tracking technology since we have to first get the full copy of the virtual machine across. However, organizations can “seed” the virtual machine in the replicated site first before starting the process of the iterative changed blocks being copied across the WAN. Vembu BDR Suite allows organizations to “seed” data in the offsite copy location first before beginning the copy operations.
Leveraging these great backup solution features can help to greatly alleviate bandwidth and network contention when thinking about replication as well as backup copy jobs across the WAN.
Network Addressing
Crucially important in thinking about failing over to a replicated DR site in a true DR situation is thinking about network addressing. Generally speaking different sites use different network address ranges. If virtual machines have been replicated from production to a DR facility, failover would require those virtual machines to be readdressed to match the local subnet of the DR facility network.
Vembu BDR Suite allows for Network & IP-ReMapping options in the configuration of VM replication. This allows for the automating of re-IP’ing virtual machines that are powered on in the replicated site vs. the production site which certainly takes the heavy lifting out of manually adjusting network settings on virtual machines during a failover.
Organizations still must take into account DNS and how DNS may get updated during a true failover situation. Are legacy applications able to deal with the change in IP address of a server resource? Although hard coding IP addresses is certainly not best practice, organizations need to make sure that applications look for resources based on DNS name rather than a hard-coded IP address.
Thoughts
Network considerations are certainly an integral part of data protection design strategies. Organizations that fail to take the network into consideration in their data protection strategy are setting themselves up for failure especially in a failover situation. Such network concerns as bandwidth contention, types of backups, and network addressing all play a part in successfully protecting production workloads from failure while minimizing strain on production networks. Using powerful backup and replication tools provided by backup solutions like Vembu BDR Suite can greatly alleviate or offset many of the network concerns that are inherent in today’s hybrid networks.
Follow our Twitter and Facebook feeds for new releases, updates, insightful posts and more.
