2012 in review: The “Living backup” is now the IT standard
Affirmed Systems is a real-time managed services provider that monitors, detects and alerts on customer performance metrics 24 hours a day, 7 days a week.
The history of backups
Data backups for many years have been static, or monolithic copies of production systems taken at scheduled intervals. Often backups were stored on magnetic tape for easy offsite archival, or on dedicated appliances. In some more advanced environments, storage area network (SAN) systems perform backups between 1 or more SAN disk arrays, either within a single site or between multiple locations to achieve disaster recovery and data protection.
These backups did protect the organization from complete loss of data in the event of disaster such as fire or flood. However these backups were NOT very good at avoiding downtime due to less permanent disasters - power or connectivity outages, server crashes, raid card failures - all things that routinely took production systems offline and effected business operations.
Redundancy solutions or products that complement a backup
In addition to doing backups, many firms deployed products like DoubleTake and could replicate live systems to hot or warm standby systems. Redundancy solutions native to the operating system or application became mature - such as Exchange 2010 Database Availability Groups and Windows SQL Clustering. Coupled with global server load balancing technologies and dns change mechanisms - the “server farm” could be highly redundant and available even in the event a complete site went offline. The challenge of these products was IT staff would have to train and document procedures for bringing the disaster recovery site online and synchronized to run as the production environment. The complexity and costs of building and maintaining these redundancies also kept their increased up-time gains out of the reach of most small to medium sized businesses.
Combining backups with redundancy to create the “Living Backup”
Our Managed Services customers often ask us when we integrate a new or existing environment in our platform - “What type of backups do you do?”
Our answer to the customer is “Living Backups”. A Living Backup is backing up a server or filesystem as a replica that is deployed directly into a virtual cloud platform. The backup is a live, up to date copy of the entire system being backed up. At each interval, the backups re-seed and update the replicas. This provides the customer a ready to go, warm standby system with optional geographic diversity from their primary server facility or office.
The backup is now no longer static, or monolithic but completely useable in a self-contained format offsite. Software products from Acronis, Appassure and others allow production physical or virtual servers including all the live data to be replicated off-site into a private or public cloud, ran as live servers, then replicated back easily when the primary site or server is restored. This technology is affordable and can be setup for a small fraction of what this type of redundancy would have cost 5 to 7 years ago in DoubleTake or using a SAN to facilitate data replication.
The Living Backup strategy is also very secure and through version control, 2 to many revisions of each snapshot backup can be kept available in the cloud or moved to other online or offline backup media facilities by the Managed Services Provider.
Current trends and how to capitalize on them
Private and public clouds for running Living Backups are now affordable and within reach. Many small businesses run software and database systems that must run on extremely fast, high performance physical servers, with SSD raid or Fusion-IO based storage systems. These systems are proprietary, generate tons of revenue and can not use public cloud services such as Amazon, Rackspace, Google apps etc. These systems must also have tremendous levels of backup and resiliency - they will need to be available at all times and under all conditions - regardless of hardware or network failures.
In 2012 we converted our last few tape backup customers to fully redundant offsite virtual systems on our private cloud. Our standard customer production deployment is vmware esxi on the bare metal with Windows or Linux guest servers. After testing many backup products for vmware esxi, we standardized on Acronis vmProtect 8. This product utilizes Changed Block Tracking (CBT) technology to backup changes using virtual disk file delta’s. After an initial full backup, only the virtual disk delta’s are then backed up to our cloud. Acronis vmprotect “updates” the replica using a delta merge process to update the replica’s virtual disk, which contains the entire operating system and all production data. This process happens without effecting the production servers as frequent as every 30 minutes. We tested even more frequent replication, but so far 30 minutes has been a good fit for our customers. For customers who do not yet use vmware in their production environment, we utilize Appassure for doing Living Backups of Windows servers. For Linux, we use proprietary internal scripts and packages to keep replicas continuously updated from production to our cloud.
Lessons learned from Hurricane Sandy
Hurricane Sandy struck New York City and took dozens of public data centers offline in Manhattan and the surrounding area for days. Companies large and small as well as government agency server farms, many inside offices were completely offline due to power and Internet outages. Many data center facilities that had backup generators found issues with their generators once running on them and lost power multiple times. This took systems offline repeatedly, corrupted data and damaged components. It was a painful and expensive lesson many firms are still struggling to recover from.
During the storm and immediately after, our team swung our Living Backups into production - by changing dns records of publicly available customer systems to our cloud. For customers working from home, internal systems were available at our facilities via backup VPN and remote desktop services. Some of our customers were up moments after Sandy disrupted power service in Manhattan, already online in our remote data center locations. Some customers required special procedures we had not planned for unique to their organization. We worked around the clock to restore production services - starting even before the storm had left the area. In the coming months, our procedures will become even more defined, adding more standardization and policies to prepare in advance for supporting proprietary as well as common systems. While there is often no “one size fits all” approach to Cloud backup and availability - testing and reviewing backup reports multiple times per hour and per day insures customer data is always available where and when it needs to be.
Thank you to all our customers for making 2012 a great year. As always, its a responsibility and honor we live by to make sure your systems and data are safe and secure.