Getting data offsite is a requirement to meet compliance directives or SLAs for many companies and cloud backup is an option that makes sense. By putting the ultimate storage target in the cloud, a data protection application can kill two birds with one stone - backup and disaster recovery (DR). The question is how to best implement this capable technology, given the existing infrastructure.

The conventional wisdom is to use an on-site appliance. This can provide a central backup target with a built-in transfer to the cloud. But do you need an on-site appliance to get the cloud’s benefits added to your backups? And is the backup appliance the best approach for all client types?

What is a Backup Appliance?

The backup appliance in this discussion is a storage device that’s installed in the data center and serves as the on-site target for backup jobs, providing a local storage repository for these data sets and an interface to the ultimate cloud storage destination. Most also include a backup application, which replaces the existing software running on backup clients, some using a lightweight piece of ‘agent’ software, others being ‘agent-less’.

Cloud backup appliances handle the backend data transfer to and from the cloud, which can be a data center operated by the appliance vendor or a public cloud. This includes supplying processing power for data reduction (typically deduplication and compression), replication over the WAN and any protocol translation that’s required to interface with the cloud.

An on-site backup appliance is a popular way to implement a cloud backup solution, but it’s not the only way, and it’s not always the best way. In many instances it makes more sense to install a lightweight agent on client computers and servers, bypass the appliance and go directly to the cloud destination.

If it “ain’t broke”

For companies that already have a backup process that’s working well, why rip and replace it just to get their data into the cloud? It’s true that backup systems have evolved significantly over the past decade or so, moving from a norm of on-site backup servers with direct-attached tape drives to network attached disk appliances with deduplication, and now ‘hybrid’ infrastructures that connect to the cloud. But most of these developments have been incremental, adding new technologies to existing systems, not replacing them altogether.

The first disk backup systems were designed to work with software that was written for tape storage and while some dedupe appliances include their own backup applications, most can work as a storage target for the in-place backup software. This same rationale applies to the cloud. For many users the best solution may be to add an agent that runs along side their current backup software on their client computers that are already backing up to a local backup server, then start putting the most critical of these backups off to the cloud. This combined solution gives the best of both worlds without replacing what is already working. Backups continue to be processed locally but a copy is in the cloud “just in case”

Choke point

A backup appliance funnels all backups into a single box, which can require scheduling and a certain amount of load balancing in order to complete backups within prescribed windows. Consolidation in IT is generally a good thing but it does add complexity and can create a single point of failure. Running backups directly to the cloud from client servers and computers is a simple solution that gets data offsite, while eliminating a potential choke point.

More CPU?

At one point, implementing new applications, like centralized backup, through appliances was an effective way to keep from burdening existing servers with another processing load. But with CPU power now abundant in most environments this is no longer a reason to buy an appliance. Most client desktops, laptops and servers have plenty of processing power to handle data transfers to the cloud while continuing with their normal workloads. Solutions like Zetta DataProtect, which installs as a lightweight agent on a client machine and maintains a synchronized copy of protected folders in the cloud, are ideal solutions for many environments.

Return of the agents

Another variant on the CPU discussion is the question of using an agent or going ‘agent-less’. Backup agents at one time consumed processing power on the computers they were protecting, potentially impacting other applications. But now, with more powerful CPUs the agent-based architecture has returned to favor. A client-side agent can support more functionality, like application awareness to improve efficiency of handling specific data types and historic snapshot-based recoveries. Agents can also be a better choice for environments with limited bandwidth or frequently disconnected mobile devices, since they don’t rely on “push” updates, like agent-less technologies do, which require regular network connectivity. 

More restore options

A client-side cloud backup solution enables individual users to access their backed up files in the cloud from a web browser and restore only the files or folders they need - and only the versions they need. They can even mount the recovery volume on their desktop. Restores can be made to dissimilar (less expensive) hardware as well, helping to keep costs down. Even iPads can access the recovery volume and its data as long as they have an application to execute that data.

Mobile users

Most users now work on notebooks (or pad computers) and use these devices to take work home and on the road. An agent-based, on-line backup solution stays with each of these computers providing data protection from anywhere. It can be the primary backup solution or can provide remote data protection when a user is traveling and can’t get access to an appliance in the office or company data center.

A primary data protection alternative   

While it can augment an existing backup infrastructure with cloud capabilities, a client-side backup application, like Zetta’s DataProtect, can provide a viable alternative to a traditional backup infrastructure as well. An agent can be installed and configured in a  few minutes and start taking backups of unstructured data, databases like SQL and Exchange, even VMware hosts. And, all server and laptops backup clients, both local and remote, can be managed from a single web-based console.

Summary

Onsite backup appliances are a useful option for many cloud backup implementations. But they’re by no means the only option. As an example, for companies that just want to add cloud DR to their existing backup system, an appliance which comes with its own backup software may be overkill. And, with the advent of more powerful CPUs and less expensive bandwidth a local backup appliance may not be needed to get data into the cloud efficiently. By providing more restore options, better support for mobile devices and a simpler implementation, client-side agents running backups directly to the cloud are often the right solution.

Zetta is a client of Storage Switzerland

Previous Entry: “Rack Density - The Key to Maximum Virtualization ROI”