Criteria for vetting appliance-based data backup systems

How does the vendor deliver management and operations features?

Installing an integrated data backup appliance offloads many infrastructure management tasks to the vendor. It is important to understand how these features are delivered. Ask the following questions:

• Does the appliance have specific physical characteristics, such as rack size, power or cooling needs?
• How does the data backup appliance manage upgrades and patches? Are they done remotely, and does the vendor need network or internet access to the appliance?
• How is the appliance locked down, and how does it match up with my organization's internal security policies? Specifically, ask about open ports and use of privileged logins.
• How are initial and additional nodes deployed in scale-out products? Do I need to pay for professional services? Is this included in the cost of the product, or can a local system administrator do the work?

From an internal management perspective, it is important to clarify if agents are needed to run the appliance. Some platforms may need client backup agents to operate, which may have to be upgraded in line with the appliance itself.

Determine your organization's scalability requirements

The requirements for scalable backup typically fall into two areas: the ability to scale capacity and the ability to scale performance. Capacity scalability means being able to support a large volume of backups and to add to that capacity on demand. Some appliance-based data backup systems do this in a scale-out manner, while others scale up.

Scale-up means that the vendor offers multiple sizes of appliances, with greater disk capacities or drive counts. Scaling in this instance might mean adding more capacity to an existing system or replacing an existing system with a new one of higher capacity.

Scale-out products increase capacity through the addition of nodes to a configuration. Some products work as a closely integrated cluster, where extra nodes extend the cluster configuration. Other integrated data backup systems treat extra nodes as part of a federated or loosely coupled cluster. Federated products simply manage multiple nodes through a single interface but do not provide resiliency across nodes. This can affect the other component of scalability: performance.

Scaling performance means increasing the bandwidth of the backup product to handle more concurrent backups or a greater amount of data throughput, which is typically measured in megabytes per second or gigabytes per second. It is important to take multiple concurrent backups in environments where a fixed backup window exists. As more physical or virtual servers are deployed, the number of concurrent backup streams will increase.

Common sense says that scale-out products are likely to provide better performance scalability, as each extra node can take part in handling backups. Each node adds networking, internal bandwidth, processor power and disk capacity to a configuration. However, be aware that the backup sources may not be able to stream as effectively as the backup system can scale, so simply increasing the size of a backup cluster may not be the answer to a performance problem.

What is your organization's architecture?

The integrated data backup appliance can manage the topography of your environment as part of the backup design. Some organizations may have a centralized environment with one or two data centers, whereas others may be highly distributed, running many remote or branch office systems.

Appliance-based data backup systems can manage both scenarios, either in the federated or clustered models discussed above. Some products are available as virtual appliances, enabling them to be integrated into a highly virtualized footprint, such as a branch office. In cases where this feature is offered, it is important to ensure that all backups are managed under a single catalogue. This enables, for example, branch data to be restored in a core or other location in the event of a hardware failure or disaster.

How reliable is the vendor's integrated data backup appliance?

Appliances are disk-based, so your organization will need to implement RAID or erasure coding to cope with hardware failure. Scale-out systems will distribute data across multiple nodes for redundancy. However, a single scale-up data backup appliance will depend on local data protection and multiple controller architecture. Here you should ask the following questions:

• How does a node failure or controller failure impact performance?
• How resilient is the backup system? Will it achieve five nines?
• How is data distributed and rebuilt across multiple nodes to facilitate recovery from failures?

If the data backup appliance is also being used for other secondary storage, such as test and development, availability can become a critical factor.

What are your organization's recovery service level needs?

Recovery time objectives (RTOs) and recovery point objectives (RPOs) define recovery service levels. RTO determines how quickly and RPO determines how far back a backup can be recovered. RTO and RPO service levels offered by a data backup appliance must be matched to the organization's expectations. You may not be able to achieve low recovery points and times with an appliance where ingestion of the backup data is slow.

Of course, in many instances, recovery is provided by a range of systems, so near-instant RTO and RPO may be provided by shared storage rather than as part of a backup.

How easy is it to replace an existing backup system?

Moving to an appliance may offer your organization an opportunity to implement new features into its backup regime.

Backup data can have a long lifetime, spanning months or years. Because of this, many IT organizations end up with multiple data backup systems and need to keep track of which system was backed up where, on a particular date and time. Ingestion of old existing backups would be ideal, although not many appliance-based backup systems offer this service.

You may also ask the vendor this question in reverse: If an appliance is deployed, how easy is it to extract backups to another platform? This second platform could be with the same vendor, where requirements justify another product option. It could also mean moving to another vendor if the product is not fit for purpose. Either way, think about the long-term implications of refreshing or replacing the backup infrastructure.

New data backup appliance features and platform support

Moving to an appliance may offer your organization an opportunity to implement new features into its backup regime. Some features that an appliance could offer over traditional backup include:

• Secondary storage enables you to use the backup platform for running virtual workloads that are not production applications. This may be for test/dev systems or simply to either test backups or restore data. If secondary usage is important, ensure that the backup platform has sufficient horsepower to meet both the needs of backup throughput and running the secondary workload at the same time.
• Search and indexing enables you to run detailed queries against the content of the backup. Integrated backup appliances are becoming more sophisticated in their search capabilities, including being able to look at files within virtual machines. Search can also be important for legal discovery and locking data from deletion.
• Compliance offers the ability to meet a range of requirements, based on industry verticals. This may mean being able to meet encryption standards for backup data or the ability to prove audit trails of backup activity.
• Archiving, or long-term preservation, enables data to be retained effectively for long periods of time, and often, this data does not change. Many vendors have added to their platforms the ability to write data to the public cloud -- typically in object stores -- and to reuse this data within the public cloud later. Cloud support does not have to be restricted to public service providers. Offering an object storage interface -- typically, Amazon S3 -- enables you to use cheaper data storage, while retaining a single metadata repository on the appliance.

Most new vendors initially focus on the server virtualization market and the simplicity of taking backups from changed block APIs. That means integrated data backup systems from new vendors often have limited platform support. Vendors then add support as their products mature. Be sure to validate that a vendor offering meets all of your application and platform needs. If some items are still on the vendor's roadmap, ask for assurances that these items will be delivered, as roadmaps change.